[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2019-05-05 Thread NedaMaleki (JIRA)


[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16833259#comment-16833259
 ] 

NedaMaleki commented on YARN-1021:
--

*Dear Wei Yan,*

*I use hadoop 2.4.1. When I want to run SLS, I face with the same problem as 
YukunTsang:*

19/05/05 11:54:45 INFO capacity.CapacityScheduler: Added node a2116.smile.com:3 
clusterResource: 
Exception in thread "main" java.lang.RuntimeException: 
java.lang.NullPointerException
    at 
org.apache.hadoop.util.ReflectionUtils.newInstance(ReflectionUtils.java:131)
    at 
org.apache.hadoop.yarn.sls.SLSRunner.startAMFromRumenTraces(SLSRunner.java:394)
    at org.apache.hadoop.yarn.sls.SLSRunner.startAM(SLSRunner.java:246)
    at org.apache.hadoop.yarn.sls.SLSRunner.start(SLSRunner.java:141)
    at org.apache.hadoop.yarn.sls.SLSRunner.main(SLSRunner.java:524)
Caused by: java.lang.NullPointerException
    at java.util.concurrent.ConcurrentHashMap.get(ConcurrentHashMap.java:936)
    at 
org.apache.hadoop.util.ReflectionUtils.newInstance(ReflectionUtils.java:123)
    ... 4 more

*After waiting some minutes I got the following messages and then nothing :(*

19/05/05 12:06:03 INFO util.AbstractLivelinessMonitor: 
Expired:a2115.smile.com:0 Timed out after 600 secs
19/05/05 12:06:03 INFO util.AbstractLivelinessMonitor: 
Expired:a2118.smile.com:1 Timed out after 600 secs
19/05/05 12:06:03 INFO util.AbstractLivelinessMonitor: 
Expired:a2117.smile.com:2 Timed out after 600 secs
19/05/05 12:06:03 INFO util.AbstractLivelinessMonitor: 
Expired:a2116.smile.com:3 Timed out after 600 secs
19/05/05 12:06:03 INFO rmnode.RMNodeImpl: Deactivating Node a2115.smile.com:0 
as it is now LOST
19/05/05 12:06:03 INFO rmnode.RMNodeImpl: a2115.smile.com:0 Node Transitioned 
from RUNNING to LOST
19/05/05 12:06:03 INFO rmnode.RMNodeImpl: Deactivating Node a2118.smile.com:1 
as it is now LOST
19/05/05 12:06:03 INFO rmnode.RMNodeImpl: a2118.smile.com:1 Node Transitioned 
from RUNNING to LOST
19/05/05 12:06:03 INFO rmnode.RMNodeImpl: Deactivating Node a2117.smile.com:2 
as it is now LOST
19/05/05 12:06:03 INFO rmnode.RMNodeImpl: a2117.smile.com:2 Node Transitioned 
from RUNNING to LOST
19/05/05 12:06:03 INFO rmnode.RMNodeImpl: Deactivating Node a2116.smile.com:3 
as it is now LOST
19/05/05 12:06:03 INFO rmnode.RMNodeImpl: a2116.smile.com:3 Node Transitioned 
from RUNNING to LOST
19/05/05 12:06:03 INFO capacity.CapacityScheduler: Removed node 
a2115.smile.com:0 clusterResource: 
19/05/05 12:06:03 INFO capacity.CapacityScheduler: Removed node 
a2118.smile.com:1 clusterResource: 
19/05/05 12:06:03 INFO capacity.CapacityScheduler: Removed node 
a2117.smile.com:2 clusterResource: 
19/05/05 12:06:03 INFO capacity.CapacityScheduler: Removed node 
a2116.smile.com:3 clusterResource: 

*I noticed when it reaches to , it shoots the 
exception and I do not know why.*

 *1) I am looking forward to hear from you as I stuck here!*

*2) My second question is that, where I can extend SLS i.e. where shall I write 
my scheduler code in SLS, run it, and get results? (I need to simulate my 
scheduler and then compare it with other schedulers like FIFO, Fair, and 
Capacity)*

*Thanks a lot,*

 *Neda*

> Yarn Scheduler Load Simulator
> -
>
> Key: YARN-1021
> URL: https://issues.apache.org/jira/browse/YARN-1021
> Project: Hadoop YARN
>  Issue Type: New Feature
>  Components: scheduler
>Reporter: Wei Yan
>Assignee: Wei Yan
>Priority: Major
> Fix For: 2.3.0
>
> Attachments: YARN-1021-demo.tar.gz, YARN-1021-images.tar.gz, 
> YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
> YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
> YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
> YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.pdf
>
>
> The Yarn Scheduler is a fertile area of interest with different 
> implementations, e.g., Fifo, Capacity and Fair  schedulers. Meanwhile, 
> several optimizations are also made to improve scheduler performance for 
> different scenarios and workload. Each scheduler algorithm has its own set of 
> features, and drives scheduling decisions by many factors, such as fairness, 
> capacity guarantee, resource availability, etc. It is very important to 
> evaluate a scheduler algorithm very well before we deploy it in a production 
> cluster. Unfortunately, currently it is non-trivial to evaluate a scheduling 
> algorithm. Evaluating in a real cluster is always time and cost consuming, 
> and it is also very hard to find a large-enough cluster. Hence, a simulator 
> which can predict how well a scheduler algorithm for some specific workload 
> would be quite useful.
> We want to build a Scheduler Load Simulator to simulate large-scale Yarn 
> clusters and application loads in a single machine. 

[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2015-10-11 Thread YukunTsang (JIRA)

[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=14952216#comment-14952216
 ] 

YukunTsang commented on YARN-1021:
--

Hey Wei Yan, I tried to run this simulator on my YARN cluster, but some problem 
occured when I am running the simulator using command "bin/slsrun.sh 
--input-rumen=sample-data/2jobs2min-rumen-jh.json --output-dir=output/"

The logs I get from the cmd are as follows:
Exception in thread "main" java.lang.RuntimeException: 
java.lang.NullPointerException
at 
org.apache.hadoop.util.ReflectionUtils.newInstance(ReflectionUtils.java:134)
at 
org.apache.hadoop.yarn.sls.SLSRunner.startAMFromRumenTraces(SLSRunner.java:398)
at org.apache.hadoop.yarn.sls.SLSRunner.startAM(SLSRunner.java:250)
at org.apache.hadoop.yarn.sls.SLSRunner.start(SLSRunner.java:145)
at org.apache.hadoop.yarn.sls.SLSRunner.main(SLSRunner.java:528)
Caused by: java.lang.NullPointerException
at 
java.util.concurrent.ConcurrentHashMap.get(ConcurrentHashMap.java:936)
at 
org.apache.hadoop.util.ReflectionUtils.newInstance(ReflectionUtils.java:126)
... 4 more

What should I do to get rid of this problem?
P.S. I used the default configuration file

> Yarn Scheduler Load Simulator
> -
>
> Key: YARN-1021
> URL: https://issues.apache.org/jira/browse/YARN-1021
> Project: Hadoop YARN
>  Issue Type: New Feature
>  Components: scheduler
>Reporter: Wei Yan
>Assignee: Wei Yan
> Fix For: 2.3.0
>
> Attachments: YARN-1021-demo.tar.gz, YARN-1021-images.tar.gz, 
> YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
> YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
> YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
> YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.pdf
>
>
> The Yarn Scheduler is a fertile area of interest with different 
> implementations, e.g., Fifo, Capacity and Fair  schedulers. Meanwhile, 
> several optimizations are also made to improve scheduler performance for 
> different scenarios and workload. Each scheduler algorithm has its own set of 
> features, and drives scheduling decisions by many factors, such as fairness, 
> capacity guarantee, resource availability, etc. It is very important to 
> evaluate a scheduler algorithm very well before we deploy it in a production 
> cluster. Unfortunately, currently it is non-trivial to evaluate a scheduling 
> algorithm. Evaluating in a real cluster is always time and cost consuming, 
> and it is also very hard to find a large-enough cluster. Hence, a simulator 
> which can predict how well a scheduler algorithm for some specific workload 
> would be quite useful.
> We want to build a Scheduler Load Simulator to simulate large-scale Yarn 
> clusters and application loads in a single machine. This would be invaluable 
> in furthering Yarn by providing a tool for researchers and developers to 
> prototype new scheduler features and predict their behavior and performance 
> with reasonable amount of confidence, there-by aiding rapid innovation.
> The simulator will exercise the real Yarn ResourceManager removing the 
> network factor by simulating NodeManagers and ApplicationMasters via handling 
> and dispatching NM/AMs heartbeat events from within the same JVM.
> To keep tracking of scheduler behavior and performance, a scheduler wrapper 
> will wrap the real scheduler.
> The simulator will produce real time metrics while executing, including:
> * Resource usages for whole cluster and each queue, which can be utilized to 
> configure cluster and queue's capacity.
> * The detailed application execution trace (recorded in relation to simulated 
> time), which can be analyzed to understand/validate the  scheduler behavior 
> (individual jobs turn around time, throughput, fairness, capacity guarantee, 
> etc).
> * Several key metrics of scheduler algorithm, such as time cost of each 
> scheduler operation (allocate, handle, etc), which can be utilized by Hadoop 
> developers to find the code spots and scalability limits.
> The simulator will provide real time charts showing the behavior of the 
> scheduler and its performance.
> A short demo is available http://www.youtube.com/watch?v=6thLi8q0qLE, showing 
> how to use simulator to simulate Fair Scheduler and Capacity Scheduler.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)


[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2015-04-24 Thread Harmandeep Singh Pall (JIRA)

[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14510697#comment-14510697
 ] 

Harmandeep Singh Pall commented on YARN-1021:
-

Hey [~ywskycn], I am trying to run the sls with the sample configurations and 
data, as suggested here but I get the same null pointer exception.

15/04/24 05:18:58 INFO rmnode.RMNodeImpl: a2115.smile.com:3 Node Transitioned 
from NEW to RUNNING
15/04/24 05:18:58 INFO fair.FairScheduler: Added node a2115.smile.com:3 cluster 
capacity: memory:40960, vCores:40
Exception in thread main java.lang.RuntimeException: 
java.lang.NullPointerException
at 
org.apache.hadoop.util.ReflectionUtils.newInstance(ReflectionUtils.java:131)
at 
org.apache.hadoop.yarn.sls.SLSRunner.startAMFromRumenTraces(SLSRunner.java:398)
at org.apache.hadoop.yarn.sls.SLSRunner.startAM(SLSRunner.java:250)
at org.apache.hadoop.yarn.sls.SLSRunner.start(SLSRunner.java:145)
at org.apache.hadoop.yarn.sls.SLSRunner.main(SLSRunner.java:528)
Caused by: java.lang.NullPointerException
at 
java.util.concurrent.ConcurrentHashMap.hash(ConcurrentHashMap.java:333)
at 
java.util.concurrent.ConcurrentHashMap.get(ConcurrentHashMap.java:988)
at 
org.apache.hadoop.util.ReflectionUtils.newInstance(ReflectionUtils.java:123)
... 4 more

Your help would be highly appreciated.

 Yarn Scheduler Load Simulator
 -

 Key: YARN-1021
 URL: https://issues.apache.org/jira/browse/YARN-1021
 Project: Hadoop YARN
  Issue Type: New Feature
  Components: scheduler
Reporter: Wei Yan
Assignee: Wei Yan
 Fix For: 2.3.0

 Attachments: YARN-1021-demo.tar.gz, YARN-1021-images.tar.gz, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.pdf


 The Yarn Scheduler is a fertile area of interest with different 
 implementations, e.g., Fifo, Capacity and Fair  schedulers. Meanwhile, 
 several optimizations are also made to improve scheduler performance for 
 different scenarios and workload. Each scheduler algorithm has its own set of 
 features, and drives scheduling decisions by many factors, such as fairness, 
 capacity guarantee, resource availability, etc. It is very important to 
 evaluate a scheduler algorithm very well before we deploy it in a production 
 cluster. Unfortunately, currently it is non-trivial to evaluate a scheduling 
 algorithm. Evaluating in a real cluster is always time and cost consuming, 
 and it is also very hard to find a large-enough cluster. Hence, a simulator 
 which can predict how well a scheduler algorithm for some specific workload 
 would be quite useful.
 We want to build a Scheduler Load Simulator to simulate large-scale Yarn 
 clusters and application loads in a single machine. This would be invaluable 
 in furthering Yarn by providing a tool for researchers and developers to 
 prototype new scheduler features and predict their behavior and performance 
 with reasonable amount of confidence, there-by aiding rapid innovation.
 The simulator will exercise the real Yarn ResourceManager removing the 
 network factor by simulating NodeManagers and ApplicationMasters via handling 
 and dispatching NM/AMs heartbeat events from within the same JVM.
 To keep tracking of scheduler behavior and performance, a scheduler wrapper 
 will wrap the real scheduler.
 The simulator will produce real time metrics while executing, including:
 * Resource usages for whole cluster and each queue, which can be utilized to 
 configure cluster and queue's capacity.
 * The detailed application execution trace (recorded in relation to simulated 
 time), which can be analyzed to understand/validate the  scheduler behavior 
 (individual jobs turn around time, throughput, fairness, capacity guarantee, 
 etc).
 * Several key metrics of scheduler algorithm, such as time cost of each 
 scheduler operation (allocate, handle, etc), which can be utilized by Hadoop 
 developers to find the code spots and scalability limits.
 The simulator will provide real time charts showing the behavior of the 
 scheduler and its performance.
 A short demo is available http://www.youtube.com/watch?v=6thLi8q0qLE, showing 
 how to use simulator to simulate Fair Scheduler and Capacity Scheduler.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)


[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2015-04-24 Thread Wei Yan (JIRA)

[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14511246#comment-14511246
 ] 

Wei Yan commented on YARN-1021:
---

[~harman], could u send me an email including your detail steps? BTW, YARN-1393 
updated a quick-start tutorial. 

 Yarn Scheduler Load Simulator
 -

 Key: YARN-1021
 URL: https://issues.apache.org/jira/browse/YARN-1021
 Project: Hadoop YARN
  Issue Type: New Feature
  Components: scheduler
Reporter: Wei Yan
Assignee: Wei Yan
 Fix For: 2.3.0

 Attachments: YARN-1021-demo.tar.gz, YARN-1021-images.tar.gz, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.pdf


 The Yarn Scheduler is a fertile area of interest with different 
 implementations, e.g., Fifo, Capacity and Fair  schedulers. Meanwhile, 
 several optimizations are also made to improve scheduler performance for 
 different scenarios and workload. Each scheduler algorithm has its own set of 
 features, and drives scheduling decisions by many factors, such as fairness, 
 capacity guarantee, resource availability, etc. It is very important to 
 evaluate a scheduler algorithm very well before we deploy it in a production 
 cluster. Unfortunately, currently it is non-trivial to evaluate a scheduling 
 algorithm. Evaluating in a real cluster is always time and cost consuming, 
 and it is also very hard to find a large-enough cluster. Hence, a simulator 
 which can predict how well a scheduler algorithm for some specific workload 
 would be quite useful.
 We want to build a Scheduler Load Simulator to simulate large-scale Yarn 
 clusters and application loads in a single machine. This would be invaluable 
 in furthering Yarn by providing a tool for researchers and developers to 
 prototype new scheduler features and predict their behavior and performance 
 with reasonable amount of confidence, there-by aiding rapid innovation.
 The simulator will exercise the real Yarn ResourceManager removing the 
 network factor by simulating NodeManagers and ApplicationMasters via handling 
 and dispatching NM/AMs heartbeat events from within the same JVM.
 To keep tracking of scheduler behavior and performance, a scheduler wrapper 
 will wrap the real scheduler.
 The simulator will produce real time metrics while executing, including:
 * Resource usages for whole cluster and each queue, which can be utilized to 
 configure cluster and queue's capacity.
 * The detailed application execution trace (recorded in relation to simulated 
 time), which can be analyzed to understand/validate the  scheduler behavior 
 (individual jobs turn around time, throughput, fairness, capacity guarantee, 
 etc).
 * Several key metrics of scheduler algorithm, such as time cost of each 
 scheduler operation (allocate, handle, etc), which can be utilized by Hadoop 
 developers to find the code spots and scalability limits.
 The simulator will provide real time charts showing the behavior of the 
 scheduler and its performance.
 A short demo is available http://www.youtube.com/watch?v=6thLi8q0qLE, showing 
 how to use simulator to simulate Fair Scheduler and Capacity Scheduler.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)


[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2015-01-20 Thread Wei Yan (JIRA)

[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14284099#comment-14284099
 ] 

Wei Yan commented on YARN-1021:
---

Thanks for the reply, [~kasha].

[~fcolzada], sorry for the late reply. I just checked hadoop-2.6.0 using the 
sample-conf and sample-data, it works fine.
The first exception is because the web module not loaded yet. Just need to wait 
2-3 seconds after you start the simulator. This exception would not affect the 
simulator running.
For the second one, could u send you config and workload files to me? I can 
look into it.

[~kasha], could u help to review YARN-1393, which provides a quick-start 
tutorial.

 Yarn Scheduler Load Simulator
 -

 Key: YARN-1021
 URL: https://issues.apache.org/jira/browse/YARN-1021
 Project: Hadoop YARN
  Issue Type: New Feature
  Components: scheduler
Reporter: Wei Yan
Assignee: Wei Yan
 Fix For: 2.3.0

 Attachments: YARN-1021-demo.tar.gz, YARN-1021-images.tar.gz, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.pdf


 The Yarn Scheduler is a fertile area of interest with different 
 implementations, e.g., Fifo, Capacity and Fair  schedulers. Meanwhile, 
 several optimizations are also made to improve scheduler performance for 
 different scenarios and workload. Each scheduler algorithm has its own set of 
 features, and drives scheduling decisions by many factors, such as fairness, 
 capacity guarantee, resource availability, etc. It is very important to 
 evaluate a scheduler algorithm very well before we deploy it in a production 
 cluster. Unfortunately, currently it is non-trivial to evaluate a scheduling 
 algorithm. Evaluating in a real cluster is always time and cost consuming, 
 and it is also very hard to find a large-enough cluster. Hence, a simulator 
 which can predict how well a scheduler algorithm for some specific workload 
 would be quite useful.
 We want to build a Scheduler Load Simulator to simulate large-scale Yarn 
 clusters and application loads in a single machine. This would be invaluable 
 in furthering Yarn by providing a tool for researchers and developers to 
 prototype new scheduler features and predict their behavior and performance 
 with reasonable amount of confidence, there-by aiding rapid innovation.
 The simulator will exercise the real Yarn ResourceManager removing the 
 network factor by simulating NodeManagers and ApplicationMasters via handling 
 and dispatching NM/AMs heartbeat events from within the same JVM.
 To keep tracking of scheduler behavior and performance, a scheduler wrapper 
 will wrap the real scheduler.
 The simulator will produce real time metrics while executing, including:
 * Resource usages for whole cluster and each queue, which can be utilized to 
 configure cluster and queue's capacity.
 * The detailed application execution trace (recorded in relation to simulated 
 time), which can be analyzed to understand/validate the  scheduler behavior 
 (individual jobs turn around time, throughput, fairness, capacity guarantee, 
 etc).
 * Several key metrics of scheduler algorithm, such as time cost of each 
 scheduler operation (allocate, handle, etc), which can be utilized by Hadoop 
 developers to find the code spots and scalability limits.
 The simulator will provide real time charts showing the behavior of the 
 scheduler and its performance.
 A short demo is available http://www.youtube.com/watch?v=6thLi8q0qLE, showing 
 how to use simulator to simulate Fair Scheduler and Capacity Scheduler.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)


[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2015-01-19 Thread Karthik Kambatla (JIRA)

[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14282469#comment-14282469
 ] 

Karthik Kambatla commented on YARN-1021:


[~fcolzada] - since this JIRA has been long committed and part of a release, do 
you mind filing a new JIRA with Affects Version 2.6.0, so it can be fixed. 
Also, if you have a patch for the issue, I ll be more than happy to help get it 
in. 

 Yarn Scheduler Load Simulator
 -

 Key: YARN-1021
 URL: https://issues.apache.org/jira/browse/YARN-1021
 Project: Hadoop YARN
  Issue Type: New Feature
  Components: scheduler
Reporter: Wei Yan
Assignee: Wei Yan
 Fix For: 2.3.0

 Attachments: YARN-1021-demo.tar.gz, YARN-1021-images.tar.gz, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.pdf


 The Yarn Scheduler is a fertile area of interest with different 
 implementations, e.g., Fifo, Capacity and Fair  schedulers. Meanwhile, 
 several optimizations are also made to improve scheduler performance for 
 different scenarios and workload. Each scheduler algorithm has its own set of 
 features, and drives scheduling decisions by many factors, such as fairness, 
 capacity guarantee, resource availability, etc. It is very important to 
 evaluate a scheduler algorithm very well before we deploy it in a production 
 cluster. Unfortunately, currently it is non-trivial to evaluate a scheduling 
 algorithm. Evaluating in a real cluster is always time and cost consuming, 
 and it is also very hard to find a large-enough cluster. Hence, a simulator 
 which can predict how well a scheduler algorithm for some specific workload 
 would be quite useful.
 We want to build a Scheduler Load Simulator to simulate large-scale Yarn 
 clusters and application loads in a single machine. This would be invaluable 
 in furthering Yarn by providing a tool for researchers and developers to 
 prototype new scheduler features and predict their behavior and performance 
 with reasonable amount of confidence, there-by aiding rapid innovation.
 The simulator will exercise the real Yarn ResourceManager removing the 
 network factor by simulating NodeManagers and ApplicationMasters via handling 
 and dispatching NM/AMs heartbeat events from within the same JVM.
 To keep tracking of scheduler behavior and performance, a scheduler wrapper 
 will wrap the real scheduler.
 The simulator will produce real time metrics while executing, including:
 * Resource usages for whole cluster and each queue, which can be utilized to 
 configure cluster and queue's capacity.
 * The detailed application execution trace (recorded in relation to simulated 
 time), which can be analyzed to understand/validate the  scheduler behavior 
 (individual jobs turn around time, throughput, fairness, capacity guarantee, 
 etc).
 * Several key metrics of scheduler algorithm, such as time cost of each 
 scheduler operation (allocate, handle, etc), which can be utilized by Hadoop 
 developers to find the code spots and scalability limits.
 The simulator will provide real time charts showing the behavior of the 
 scheduler and its performance.
 A short demo is available http://www.youtube.com/watch?v=6thLi8q0qLE, showing 
 how to use simulator to simulate Fair Scheduler and Capacity Scheduler.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)


[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2015-01-19 Thread Fabio C (JIRA)

[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14282511#comment-14282511
 ] 

Fabio C commented on YARN-1021:
---

The fact is I'm not even sure it's a bug (for example I also got the same 
mysterious exceptions of [~ashoksekar07], then finding out it was just a 
missing config file). This is why I wanted to get in touch with the author 
first. If you can run sls and confirm the issue I'll go for a new issue 
submission. On the Hadoop mailing list it looks like I'm the only one using SLS.
Regards

 Yarn Scheduler Load Simulator
 -

 Key: YARN-1021
 URL: https://issues.apache.org/jira/browse/YARN-1021
 Project: Hadoop YARN
  Issue Type: New Feature
  Components: scheduler
Reporter: Wei Yan
Assignee: Wei Yan
 Fix For: 2.3.0

 Attachments: YARN-1021-demo.tar.gz, YARN-1021-images.tar.gz, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.pdf


 The Yarn Scheduler is a fertile area of interest with different 
 implementations, e.g., Fifo, Capacity and Fair  schedulers. Meanwhile, 
 several optimizations are also made to improve scheduler performance for 
 different scenarios and workload. Each scheduler algorithm has its own set of 
 features, and drives scheduling decisions by many factors, such as fairness, 
 capacity guarantee, resource availability, etc. It is very important to 
 evaluate a scheduler algorithm very well before we deploy it in a production 
 cluster. Unfortunately, currently it is non-trivial to evaluate a scheduling 
 algorithm. Evaluating in a real cluster is always time and cost consuming, 
 and it is also very hard to find a large-enough cluster. Hence, a simulator 
 which can predict how well a scheduler algorithm for some specific workload 
 would be quite useful.
 We want to build a Scheduler Load Simulator to simulate large-scale Yarn 
 clusters and application loads in a single machine. This would be invaluable 
 in furthering Yarn by providing a tool for researchers and developers to 
 prototype new scheduler features and predict their behavior and performance 
 with reasonable amount of confidence, there-by aiding rapid innovation.
 The simulator will exercise the real Yarn ResourceManager removing the 
 network factor by simulating NodeManagers and ApplicationMasters via handling 
 and dispatching NM/AMs heartbeat events from within the same JVM.
 To keep tracking of scheduler behavior and performance, a scheduler wrapper 
 will wrap the real scheduler.
 The simulator will produce real time metrics while executing, including:
 * Resource usages for whole cluster and each queue, which can be utilized to 
 configure cluster and queue's capacity.
 * The detailed application execution trace (recorded in relation to simulated 
 time), which can be analyzed to understand/validate the  scheduler behavior 
 (individual jobs turn around time, throughput, fairness, capacity guarantee, 
 etc).
 * Several key metrics of scheduler algorithm, such as time cost of each 
 scheduler operation (allocate, handle, etc), which can be utilized by Hadoop 
 developers to find the code spots and scalability limits.
 The simulator will provide real time charts showing the behavior of the 
 scheduler and its performance.
 A short demo is available http://www.youtube.com/watch?v=6thLi8q0qLE, showing 
 how to use simulator to simulate Fair Scheduler and Capacity Scheduler.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)


[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2015-01-17 Thread Fabio Colzada (JIRA)

[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14281276#comment-14281276
 ] 

Fabio Colzada commented on YARN-1021:
-

Hi, I am working with sls on Hadoop 2.6.0, I really need it but I'm struggling 
to have it running at its best. Simulation completes on command line and I can 
see the log entries on the screen, but:

1- the web interface is not working. On screen I can see this exception more or 
less at the beginning of the workflow:
java.lang.NullPointerException
at org.apache.hadoop.yarn.sls.web.SLSWebApp.init(SLSWebApp.java:86)
at 
org.apache.hadoop.yarn.sls.scheduler.ResourceSchedulerWrapper.initMetrics(ResourceSchedulerWrapper.java:477)
at 
org.apache.hadoop.yarn.sls.scheduler.ResourceSchedulerWrapper.setConf(ResourceSchedulerWrapper.java:176)
at 
org.apache.hadoop.util.ReflectionUtils.setConf(ReflectionUtils.java:73)
at 
org.apache.hadoop.util.ReflectionUtils.newInstance(ReflectionUtils.java:133)
at 
org.apache.hadoop.yarn.server.resourcemanager.ResourceManager.createScheduler
(ResourceManager.java:291)
at 
org.apache.hadoop.yarn.server.resourcemanager.ResourceManager$RMActiveServices.serviceInit
(ResourceManager.java:484)
at 
org.apache.hadoop.service.AbstractService.init(AbstractService.java:163)
at 
org.apache.hadoop.yarn.server.resourcemanager.ResourceManager.createAndInitActiveServices
(ResourceManager.java:989)
at 
org.apache.hadoop.yarn.server.resourcemanager.ResourceManager.serviceInit
(ResourceManager.java:255)
at 
org.apache.hadoop.service.AbstractService.init(AbstractService.java:163)
at org.apache.hadoop.yarn.sls.SLSRunner.startRM(SLSRunner.java:167)
at org.apache.hadoop.yarn.sls.SLSRunner.start(SLSRunner.java:141)
at org.apache.hadoop.yarn.sls.SLSRunner.main(SLSRunner.java:528)

Not sure which object is null, but I see the folder sls/html has the expected 
files.

2- I don't get the files realtimetrack.json nor jobruntime.csv, while metrics 
folder is correctly populated. I see some recurring exceptions, I don't know if 
they are related since they don't prevent the simulation to terminate:
java.lang.NullPointerException
at 
org.apache.hadoop.yarn.sls.scheduler.ResourceSchedulerWrapper.addAMRuntime
(ResourceSchedulerWrapper.java:735)
at 
org.apache.hadoop.yarn.sls.appmaster.AMSimulator.lastStep(AMSimulator.java:193)
at 
org.apache.hadoop.yarn.sls.appmaster.MRAMSimulator.lastStep(MRAMSimulator.java:396)
at 
org.apache.hadoop.yarn.sls.scheduler.TaskRunner$Task.run(TaskRunner.java:100)
at 
java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
at 
java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
at java.lang.Thread.run(Thread.java:745)

and also

Exception in thread pool-5-thread-374 java.lang.NullPointerException
at 
org.apache.hadoop.yarn.sls.scheduler.TaskRunner$Task.run(TaskRunner.java:104)
at 
java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
at 
java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
at java.lang.Thread.run(Thread.java:745)

Any help is really appreciated

 Yarn Scheduler Load Simulator
 -

 Key: YARN-1021
 URL: https://issues.apache.org/jira/browse/YARN-1021
 Project: Hadoop YARN
  Issue Type: New Feature
  Components: scheduler
Reporter: Wei Yan
Assignee: Wei Yan
 Fix For: 2.3.0

 Attachments: YARN-1021-demo.tar.gz, YARN-1021-images.tar.gz, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.pdf


 The Yarn Scheduler is a fertile area of interest with different 
 implementations, e.g., Fifo, Capacity and Fair  schedulers. Meanwhile, 
 several optimizations are also made to improve scheduler performance for 
 different scenarios and workload. Each scheduler algorithm has its own set of 
 features, and drives scheduling decisions by many factors, such as fairness, 
 capacity guarantee, resource availability, etc. It is very important to 
 evaluate a scheduler algorithm very well before we deploy it in a production 
 cluster. Unfortunately, currently it is non-trivial to evaluate a scheduling 
 algorithm. Evaluating in a real cluster is always time and cost consuming, 
 and it is also very hard to find a large-enough cluster. Hence, a simulator 
 which can predict how well a scheduler algorithm for some specific workload 
 would be quite useful.
 We want to build a Scheduler Load 

[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2014-11-07 Thread Wei Yan (JIRA)

[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14202407#comment-14202407
 ] 

Wei Yan commented on YARN-1021:
---

Hi, [~ashoksekar07], you can email your configuration files and how you start 
the simulator steps to my email. I can take a look.

 Yarn Scheduler Load Simulator
 -

 Key: YARN-1021
 URL: https://issues.apache.org/jira/browse/YARN-1021
 Project: Hadoop YARN
  Issue Type: New Feature
  Components: scheduler
Reporter: Wei Yan
Assignee: Wei Yan
 Fix For: 2.3.0

 Attachments: YARN-1021-demo.tar.gz, YARN-1021-images.tar.gz, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.pdf


 The Yarn Scheduler is a fertile area of interest with different 
 implementations, e.g., Fifo, Capacity and Fair  schedulers. Meanwhile, 
 several optimizations are also made to improve scheduler performance for 
 different scenarios and workload. Each scheduler algorithm has its own set of 
 features, and drives scheduling decisions by many factors, such as fairness, 
 capacity guarantee, resource availability, etc. It is very important to 
 evaluate a scheduler algorithm very well before we deploy it in a production 
 cluster. Unfortunately, currently it is non-trivial to evaluate a scheduling 
 algorithm. Evaluating in a real cluster is always time and cost consuming, 
 and it is also very hard to find a large-enough cluster. Hence, a simulator 
 which can predict how well a scheduler algorithm for some specific workload 
 would be quite useful.
 We want to build a Scheduler Load Simulator to simulate large-scale Yarn 
 clusters and application loads in a single machine. This would be invaluable 
 in furthering Yarn by providing a tool for researchers and developers to 
 prototype new scheduler features and predict their behavior and performance 
 with reasonable amount of confidence, there-by aiding rapid innovation.
 The simulator will exercise the real Yarn ResourceManager removing the 
 network factor by simulating NodeManagers and ApplicationMasters via handling 
 and dispatching NM/AMs heartbeat events from within the same JVM.
 To keep tracking of scheduler behavior and performance, a scheduler wrapper 
 will wrap the real scheduler.
 The simulator will produce real time metrics while executing, including:
 * Resource usages for whole cluster and each queue, which can be utilized to 
 configure cluster and queue's capacity.
 * The detailed application execution trace (recorded in relation to simulated 
 time), which can be analyzed to understand/validate the  scheduler behavior 
 (individual jobs turn around time, throughput, fairness, capacity guarantee, 
 etc).
 * Several key metrics of scheduler algorithm, such as time cost of each 
 scheduler operation (allocate, handle, etc), which can be utilized by Hadoop 
 developers to find the code spots and scalability limits.
 The simulator will provide real time charts showing the behavior of the 
 scheduler and its performance.
 A short demo is available http://www.youtube.com/watch?v=6thLi8q0qLE, showing 
 how to use simulator to simulate Fair Scheduler and Capacity Scheduler.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)


[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2014-11-05 Thread Ashok Chandrasekar (JIRA)

[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14199745#comment-14199745
 ] 

Ashok Chandrasekar commented on YARN-1021:
--

Hi Wei Yan, when I am running the SLS with 
bin/slsrun.sh --input-rumen=sample-data/2jobs2min-rumen-jh.json 
--output-dir=output/ --print-simulation
I am getting a NullPointerException as follows:
14/11/06 03:56:05 INFO rmnode.RMNodeImpl: a2115.smile.com:3 Node Transitioned 
from NEW to RUNNING
14/11/06 03:56:05 INFO capacity.CapacityScheduler: Added node a2115.smile.com:3 
clusterResource: memory:40960, vCores:40
Exception in thread main java.lang.RuntimeException: 
java.lang.NullPointerException
at 
org.apache.hadoop.util.ReflectionUtils.newInstance(ReflectionUtils.java:131)
at 
org.apache.hadoop.yarn.sls.SLSRunner.startAMFromRumenTraces(SLSRunner.java:394)
at org.apache.hadoop.yarn.sls.SLSRunner.startAM(SLSRunner.java:246)
at org.apache.hadoop.yarn.sls.SLSRunner.start(SLSRunner.java:141)
at org.apache.hadoop.yarn.sls.SLSRunner.main(SLSRunner.java:524)
Caused by: java.lang.NullPointerException
at 
java.util.concurrent.ConcurrentHashMap.hash(ConcurrentHashMap.java:333)
at 
java.util.concurrent.ConcurrentHashMap.get(ConcurrentHashMap.java:988)
at 
org.apache.hadoop.util.ReflectionUtils.newInstance(ReflectionUtils.java:123)
... 4 more

I am using a single node pseudo distributed setup. I have tried the SLS with 
both Hadoop 2.3 and 2.5, but I am still getting the same exception. Can you 
please help me identify the problem? Thank you very much.

 Yarn Scheduler Load Simulator
 -

 Key: YARN-1021
 URL: https://issues.apache.org/jira/browse/YARN-1021
 Project: Hadoop YARN
  Issue Type: New Feature
  Components: scheduler
Reporter: Wei Yan
Assignee: Wei Yan
 Fix For: 2.3.0

 Attachments: YARN-1021-demo.tar.gz, YARN-1021-images.tar.gz, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.pdf


 The Yarn Scheduler is a fertile area of interest with different 
 implementations, e.g., Fifo, Capacity and Fair  schedulers. Meanwhile, 
 several optimizations are also made to improve scheduler performance for 
 different scenarios and workload. Each scheduler algorithm has its own set of 
 features, and drives scheduling decisions by many factors, such as fairness, 
 capacity guarantee, resource availability, etc. It is very important to 
 evaluate a scheduler algorithm very well before we deploy it in a production 
 cluster. Unfortunately, currently it is non-trivial to evaluate a scheduling 
 algorithm. Evaluating in a real cluster is always time and cost consuming, 
 and it is also very hard to find a large-enough cluster. Hence, a simulator 
 which can predict how well a scheduler algorithm for some specific workload 
 would be quite useful.
 We want to build a Scheduler Load Simulator to simulate large-scale Yarn 
 clusters and application loads in a single machine. This would be invaluable 
 in furthering Yarn by providing a tool for researchers and developers to 
 prototype new scheduler features and predict their behavior and performance 
 with reasonable amount of confidence, there-by aiding rapid innovation.
 The simulator will exercise the real Yarn ResourceManager removing the 
 network factor by simulating NodeManagers and ApplicationMasters via handling 
 and dispatching NM/AMs heartbeat events from within the same JVM.
 To keep tracking of scheduler behavior and performance, a scheduler wrapper 
 will wrap the real scheduler.
 The simulator will produce real time metrics while executing, including:
 * Resource usages for whole cluster and each queue, which can be utilized to 
 configure cluster and queue's capacity.
 * The detailed application execution trace (recorded in relation to simulated 
 time), which can be analyzed to understand/validate the  scheduler behavior 
 (individual jobs turn around time, throughput, fairness, capacity guarantee, 
 etc).
 * Several key metrics of scheduler algorithm, such as time cost of each 
 scheduler operation (allocate, handle, etc), which can be utilized by Hadoop 
 developers to find the code spots and scalability limits.
 The simulator will provide real time charts showing the behavior of the 
 scheduler and its performance.
 A short demo is available http://www.youtube.com/watch?v=6thLi8q0qLE, showing 
 how to use simulator to simulate Fair Scheduler and Capacity Scheduler.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)


[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2014-08-04 Thread JiankunLiu (JIRA)

[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14084438#comment-14084438
 ] 

JiankunLiu commented on YARN-1021:
--

Hi Wei Yan,I meet with the following exception when I use this commond. 
bin/slsrun.sh --input-rumen=sample-data/2jobs2min-rumen-jh.json 
--output-dir=sample_output
Exception in thread pool-3-thread-1 java.lang.NullPointerException
at 
org.apache.hadoop.yarn.sls.appmaster.AMSimulator.registerAM(AMSimulator.java:300)
at 
org.apache.hadoop.yarn.sls.appmaster.AMSimulator.firstStep(AMSimulator.java:141)
at 
org.apache.hadoop.yarn.sls.appmaster.MRAMSimulator.firstStep(MRAMSimulator.java:146)
at 
org.apache.hadoop.yarn.sls.scheduler.TaskRunner$Task.run(TaskRunner.java:84)
at 
java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1145)
at 
java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:615)
at java.lang.Thread.run(Thread.java:744)
Can you do me a favor ? Thank you very much.

 Yarn Scheduler Load Simulator
 -

 Key: YARN-1021
 URL: https://issues.apache.org/jira/browse/YARN-1021
 Project: Hadoop YARN
  Issue Type: New Feature
  Components: scheduler
Reporter: Wei Yan
Assignee: Wei Yan
 Fix For: 2.3.0

 Attachments: YARN-1021-demo.tar.gz, YARN-1021-images.tar.gz, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.pdf


 The Yarn Scheduler is a fertile area of interest with different 
 implementations, e.g., Fifo, Capacity and Fair  schedulers. Meanwhile, 
 several optimizations are also made to improve scheduler performance for 
 different scenarios and workload. Each scheduler algorithm has its own set of 
 features, and drives scheduling decisions by many factors, such as fairness, 
 capacity guarantee, resource availability, etc. It is very important to 
 evaluate a scheduler algorithm very well before we deploy it in a production 
 cluster. Unfortunately, currently it is non-trivial to evaluate a scheduling 
 algorithm. Evaluating in a real cluster is always time and cost consuming, 
 and it is also very hard to find a large-enough cluster. Hence, a simulator 
 which can predict how well a scheduler algorithm for some specific workload 
 would be quite useful.
 We want to build a Scheduler Load Simulator to simulate large-scale Yarn 
 clusters and application loads in a single machine. This would be invaluable 
 in furthering Yarn by providing a tool for researchers and developers to 
 prototype new scheduler features and predict their behavior and performance 
 with reasonable amount of confidence, there-by aiding rapid innovation.
 The simulator will exercise the real Yarn ResourceManager removing the 
 network factor by simulating NodeManagers and ApplicationMasters via handling 
 and dispatching NM/AMs heartbeat events from within the same JVM.
 To keep tracking of scheduler behavior and performance, a scheduler wrapper 
 will wrap the real scheduler.
 The simulator will produce real time metrics while executing, including:
 * Resource usages for whole cluster and each queue, which can be utilized to 
 configure cluster and queue's capacity.
 * The detailed application execution trace (recorded in relation to simulated 
 time), which can be analyzed to understand/validate the  scheduler behavior 
 (individual jobs turn around time, throughput, fairness, capacity guarantee, 
 etc).
 * Several key metrics of scheduler algorithm, such as time cost of each 
 scheduler operation (allocate, handle, etc), which can be utilized by Hadoop 
 developers to find the code spots and scalability limits.
 The simulator will provide real time charts showing the behavior of the 
 scheduler and its performance.
 A short demo is available http://www.youtube.com/watch?v=6thLi8q0qLE, showing 
 how to use simulator to simulate Fair Scheduler and Capacity Scheduler.



--
This message was sent by Atlassian JIRA
(v6.2#6252)


[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2014-08-04 Thread JiankunLiu (JIRA)

[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14084689#comment-14084689
 ] 

JiankunLiu commented on YARN-1021:
--

I found that it was no exception on hadoop-2.3.0. But I meet with the following 
exception on hadoop-2.4.0.
Exception in thread pool-3-thread-1 java.lang.NullPointerException
at 
org.apache.hadoop.yarn.sls.appmaster.AMSimulator.registerAM(AMSimulator.java:300)
at 
org.apache.hadoop.yarn.sls.appmaster.AMSimulator.firstStep(AMSimulator.java:141)
at 
org.apache.hadoop.yarn.sls.appmaster.MRAMSimulator.firstStep(MRAMSimulator.java:146)
at org.apache.hadoop.yarn.sls.scheduler.TaskRunner$Task.run(TaskRunner.java:84)
at 
java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1145)
at 
java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:615)
at java.lang.Thread.run(Thread.java:744)
what should I do?

 Yarn Scheduler Load Simulator
 -

 Key: YARN-1021
 URL: https://issues.apache.org/jira/browse/YARN-1021
 Project: Hadoop YARN
  Issue Type: New Feature
  Components: scheduler
Reporter: Wei Yan
Assignee: Wei Yan
 Fix For: 2.3.0

 Attachments: YARN-1021-demo.tar.gz, YARN-1021-images.tar.gz, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.pdf


 The Yarn Scheduler is a fertile area of interest with different 
 implementations, e.g., Fifo, Capacity and Fair  schedulers. Meanwhile, 
 several optimizations are also made to improve scheduler performance for 
 different scenarios and workload. Each scheduler algorithm has its own set of 
 features, and drives scheduling decisions by many factors, such as fairness, 
 capacity guarantee, resource availability, etc. It is very important to 
 evaluate a scheduler algorithm very well before we deploy it in a production 
 cluster. Unfortunately, currently it is non-trivial to evaluate a scheduling 
 algorithm. Evaluating in a real cluster is always time and cost consuming, 
 and it is also very hard to find a large-enough cluster. Hence, a simulator 
 which can predict how well a scheduler algorithm for some specific workload 
 would be quite useful.
 We want to build a Scheduler Load Simulator to simulate large-scale Yarn 
 clusters and application loads in a single machine. This would be invaluable 
 in furthering Yarn by providing a tool for researchers and developers to 
 prototype new scheduler features and predict their behavior and performance 
 with reasonable amount of confidence, there-by aiding rapid innovation.
 The simulator will exercise the real Yarn ResourceManager removing the 
 network factor by simulating NodeManagers and ApplicationMasters via handling 
 and dispatching NM/AMs heartbeat events from within the same JVM.
 To keep tracking of scheduler behavior and performance, a scheduler wrapper 
 will wrap the real scheduler.
 The simulator will produce real time metrics while executing, including:
 * Resource usages for whole cluster and each queue, which can be utilized to 
 configure cluster and queue's capacity.
 * The detailed application execution trace (recorded in relation to simulated 
 time), which can be analyzed to understand/validate the  scheduler behavior 
 (individual jobs turn around time, throughput, fairness, capacity guarantee, 
 etc).
 * Several key metrics of scheduler algorithm, such as time cost of each 
 scheduler operation (allocate, handle, etc), which can be utilized by Hadoop 
 developers to find the code spots and scalability limits.
 The simulator will provide real time charts showing the behavior of the 
 scheduler and its performance.
 A short demo is available http://www.youtube.com/watch?v=6thLi8q0qLE, showing 
 how to use simulator to simulate Fair Scheduler and Capacity Scheduler.



--
This message was sent by Atlassian JIRA
(v6.2#6252)


[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2014-08-04 Thread Wei Yan (JIRA)

[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14084797#comment-14084797
 ] 

Wei Yan commented on YARN-1021:
---

[~Jackliu91], yes, the simulator has exception in hadoop-2.4.0, the patch 
provided in YARN-1726. The coming released hadoop-2.5.0 will include that patch 
and the simulator should run well.

 Yarn Scheduler Load Simulator
 -

 Key: YARN-1021
 URL: https://issues.apache.org/jira/browse/YARN-1021
 Project: Hadoop YARN
  Issue Type: New Feature
  Components: scheduler
Reporter: Wei Yan
Assignee: Wei Yan
 Fix For: 2.3.0

 Attachments: YARN-1021-demo.tar.gz, YARN-1021-images.tar.gz, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.pdf


 The Yarn Scheduler is a fertile area of interest with different 
 implementations, e.g., Fifo, Capacity and Fair  schedulers. Meanwhile, 
 several optimizations are also made to improve scheduler performance for 
 different scenarios and workload. Each scheduler algorithm has its own set of 
 features, and drives scheduling decisions by many factors, such as fairness, 
 capacity guarantee, resource availability, etc. It is very important to 
 evaluate a scheduler algorithm very well before we deploy it in a production 
 cluster. Unfortunately, currently it is non-trivial to evaluate a scheduling 
 algorithm. Evaluating in a real cluster is always time and cost consuming, 
 and it is also very hard to find a large-enough cluster. Hence, a simulator 
 which can predict how well a scheduler algorithm for some specific workload 
 would be quite useful.
 We want to build a Scheduler Load Simulator to simulate large-scale Yarn 
 clusters and application loads in a single machine. This would be invaluable 
 in furthering Yarn by providing a tool for researchers and developers to 
 prototype new scheduler features and predict their behavior and performance 
 with reasonable amount of confidence, there-by aiding rapid innovation.
 The simulator will exercise the real Yarn ResourceManager removing the 
 network factor by simulating NodeManagers and ApplicationMasters via handling 
 and dispatching NM/AMs heartbeat events from within the same JVM.
 To keep tracking of scheduler behavior and performance, a scheduler wrapper 
 will wrap the real scheduler.
 The simulator will produce real time metrics while executing, including:
 * Resource usages for whole cluster and each queue, which can be utilized to 
 configure cluster and queue's capacity.
 * The detailed application execution trace (recorded in relation to simulated 
 time), which can be analyzed to understand/validate the  scheduler behavior 
 (individual jobs turn around time, throughput, fairness, capacity guarantee, 
 etc).
 * Several key metrics of scheduler algorithm, such as time cost of each 
 scheduler operation (allocate, handle, etc), which can be utilized by Hadoop 
 developers to find the code spots and scalability limits.
 The simulator will provide real time charts showing the behavior of the 
 scheduler and its performance.
 A short demo is available http://www.youtube.com/watch?v=6thLi8q0qLE, showing 
 how to use simulator to simulate Fair Scheduler and Capacity Scheduler.



--
This message was sent by Atlassian JIRA
(v6.2#6252)


[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2014-05-27 Thread Wei Yan (JIRA)

[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14010234#comment-14010234
 ] 

Wei Yan commented on YARN-1021:
---

[~cristiana.voicu], the SLS directly supports rumen traces. In general, you 
need to have some existing workload traces (i.e., from some production 
clusters), and then use Rumen to generate workload traces. Then let the SLS 
load these traces. Or you can generate some traces randomly (random # of jobs, 
requests, lifetime, etc).
Sorry that I don't have the traces used in that page right now.

 Yarn Scheduler Load Simulator
 -

 Key: YARN-1021
 URL: https://issues.apache.org/jira/browse/YARN-1021
 Project: Hadoop YARN
  Issue Type: New Feature
  Components: scheduler
Reporter: Wei Yan
Assignee: Wei Yan
 Fix For: 2.3.0

 Attachments: YARN-1021-demo.tar.gz, YARN-1021-images.tar.gz, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.pdf


 The Yarn Scheduler is a fertile area of interest with different 
 implementations, e.g., Fifo, Capacity and Fair  schedulers. Meanwhile, 
 several optimizations are also made to improve scheduler performance for 
 different scenarios and workload. Each scheduler algorithm has its own set of 
 features, and drives scheduling decisions by many factors, such as fairness, 
 capacity guarantee, resource availability, etc. It is very important to 
 evaluate a scheduler algorithm very well before we deploy it in a production 
 cluster. Unfortunately, currently it is non-trivial to evaluate a scheduling 
 algorithm. Evaluating in a real cluster is always time and cost consuming, 
 and it is also very hard to find a large-enough cluster. Hence, a simulator 
 which can predict how well a scheduler algorithm for some specific workload 
 would be quite useful.
 We want to build a Scheduler Load Simulator to simulate large-scale Yarn 
 clusters and application loads in a single machine. This would be invaluable 
 in furthering Yarn by providing a tool for researchers and developers to 
 prototype new scheduler features and predict their behavior and performance 
 with reasonable amount of confidence, there-by aiding rapid innovation.
 The simulator will exercise the real Yarn ResourceManager removing the 
 network factor by simulating NodeManagers and ApplicationMasters via handling 
 and dispatching NM/AMs heartbeat events from within the same JVM.
 To keep tracking of scheduler behavior and performance, a scheduler wrapper 
 will wrap the real scheduler.
 The simulator will produce real time metrics while executing, including:
 * Resource usages for whole cluster and each queue, which can be utilized to 
 configure cluster and queue's capacity.
 * The detailed application execution trace (recorded in relation to simulated 
 time), which can be analyzed to understand/validate the  scheduler behavior 
 (individual jobs turn around time, throughput, fairness, capacity guarantee, 
 etc).
 * Several key metrics of scheduler algorithm, such as time cost of each 
 scheduler operation (allocate, handle, etc), which can be utilized by Hadoop 
 developers to find the code spots and scalability limits.
 The simulator will provide real time charts showing the behavior of the 
 scheduler and its performance.
 A short demo is available http://www.youtube.com/watch?v=6thLi8q0qLE, showing 
 how to use simulator to simulate Fair Scheduler and Capacity Scheduler.



--
This message was sent by Atlassian JIRA
(v6.2#6252)


[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2014-04-28 Thread Haiou Fang (JIRA)

[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13982772#comment-13982772
 ] 

Haiou Fang commented on YARN-1021:
--

Hi, [~ywskycn] I am sorry to disturb you with such question and I have run the 
sls successful. “bin/slsrun.sh --input-sls=sls-file/sls-jobs.json 
--output-dir=output_sls --nodes=sls-file/sls-nodes.json ” thank you


 Yarn Scheduler Load Simulator
 -

 Key: YARN-1021
 URL: https://issues.apache.org/jira/browse/YARN-1021
 Project: Hadoop YARN
  Issue Type: New Feature
  Components: scheduler
Reporter: Wei Yan
Assignee: Wei Yan
 Fix For: 2.3.0

 Attachments: YARN-1021-demo.tar.gz, YARN-1021-images.tar.gz, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.pdf


 The Yarn Scheduler is a fertile area of interest with different 
 implementations, e.g., Fifo, Capacity and Fair  schedulers. Meanwhile, 
 several optimizations are also made to improve scheduler performance for 
 different scenarios and workload. Each scheduler algorithm has its own set of 
 features, and drives scheduling decisions by many factors, such as fairness, 
 capacity guarantee, resource availability, etc. It is very important to 
 evaluate a scheduler algorithm very well before we deploy it in a production 
 cluster. Unfortunately, currently it is non-trivial to evaluate a scheduling 
 algorithm. Evaluating in a real cluster is always time and cost consuming, 
 and it is also very hard to find a large-enough cluster. Hence, a simulator 
 which can predict how well a scheduler algorithm for some specific workload 
 would be quite useful.
 We want to build a Scheduler Load Simulator to simulate large-scale Yarn 
 clusters and application loads in a single machine. This would be invaluable 
 in furthering Yarn by providing a tool for researchers and developers to 
 prototype new scheduler features and predict their behavior and performance 
 with reasonable amount of confidence, there-by aiding rapid innovation.
 The simulator will exercise the real Yarn ResourceManager removing the 
 network factor by simulating NodeManagers and ApplicationMasters via handling 
 and dispatching NM/AMs heartbeat events from within the same JVM.
 To keep tracking of scheduler behavior and performance, a scheduler wrapper 
 will wrap the real scheduler.
 The simulator will produce real time metrics while executing, including:
 * Resource usages for whole cluster and each queue, which can be utilized to 
 configure cluster and queue's capacity.
 * The detailed application execution trace (recorded in relation to simulated 
 time), which can be analyzed to understand/validate the  scheduler behavior 
 (individual jobs turn around time, throughput, fairness, capacity guarantee, 
 etc).
 * Several key metrics of scheduler algorithm, such as time cost of each 
 scheduler operation (allocate, handle, etc), which can be utilized by Hadoop 
 developers to find the code spots and scalability limits.
 The simulator will provide real time charts showing the behavior of the 
 scheduler and its performance.
 A short demo is available http://www.youtube.com/watch?v=6thLi8q0qLE, showing 
 how to use simulator to simulate Fair Scheduler and Capacity Scheduler.



--
This message was sent by Atlassian JIRA
(v6.2#6252)


[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2014-04-28 Thread Wei Yan (JIRA)

[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13982963#comment-13982963
 ] 

Wei Yan commented on YARN-1021:
---

[~haiouman], Sorry for late reply. Great that your problem solved.

 Yarn Scheduler Load Simulator
 -

 Key: YARN-1021
 URL: https://issues.apache.org/jira/browse/YARN-1021
 Project: Hadoop YARN
  Issue Type: New Feature
  Components: scheduler
Reporter: Wei Yan
Assignee: Wei Yan
 Fix For: 2.3.0

 Attachments: YARN-1021-demo.tar.gz, YARN-1021-images.tar.gz, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.pdf


 The Yarn Scheduler is a fertile area of interest with different 
 implementations, e.g., Fifo, Capacity and Fair  schedulers. Meanwhile, 
 several optimizations are also made to improve scheduler performance for 
 different scenarios and workload. Each scheduler algorithm has its own set of 
 features, and drives scheduling decisions by many factors, such as fairness, 
 capacity guarantee, resource availability, etc. It is very important to 
 evaluate a scheduler algorithm very well before we deploy it in a production 
 cluster. Unfortunately, currently it is non-trivial to evaluate a scheduling 
 algorithm. Evaluating in a real cluster is always time and cost consuming, 
 and it is also very hard to find a large-enough cluster. Hence, a simulator 
 which can predict how well a scheduler algorithm for some specific workload 
 would be quite useful.
 We want to build a Scheduler Load Simulator to simulate large-scale Yarn 
 clusters and application loads in a single machine. This would be invaluable 
 in furthering Yarn by providing a tool for researchers and developers to 
 prototype new scheduler features and predict their behavior and performance 
 with reasonable amount of confidence, there-by aiding rapid innovation.
 The simulator will exercise the real Yarn ResourceManager removing the 
 network factor by simulating NodeManagers and ApplicationMasters via handling 
 and dispatching NM/AMs heartbeat events from within the same JVM.
 To keep tracking of scheduler behavior and performance, a scheduler wrapper 
 will wrap the real scheduler.
 The simulator will produce real time metrics while executing, including:
 * Resource usages for whole cluster and each queue, which can be utilized to 
 configure cluster and queue's capacity.
 * The detailed application execution trace (recorded in relation to simulated 
 time), which can be analyzed to understand/validate the  scheduler behavior 
 (individual jobs turn around time, throughput, fairness, capacity guarantee, 
 etc).
 * Several key metrics of scheduler algorithm, such as time cost of each 
 scheduler operation (allocate, handle, etc), which can be utilized by Hadoop 
 developers to find the code spots and scalability limits.
 The simulator will provide real time charts showing the behavior of the 
 scheduler and its performance.
 A short demo is available http://www.youtube.com/watch?v=6thLi8q0qLE, showing 
 how to use simulator to simulate Fair Scheduler and Capacity Scheduler.



--
This message was sent by Atlassian JIRA
(v6.2#6252)


[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2014-04-27 Thread Haiou Fang (JIRA)

[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13982748#comment-13982748
 ] 

Haiou Fang commented on YARN-1021:
--

Hi, [~ywskycn]. It throwed the following exception when I have entered the 
command:  bin/slsrun.sh 
--input-sls=/.../hadoop-2.3.0/share/hadoop/tools/sls/sample-data/2jobs2min-rumen-jh.json
 --output-dir=output_exp, I really have no idea about it. Can you do me a 
favor ? Thank you very much.

Exception in thread main java.lang.NullPointerException
at 
org.apache.hadoop.yarn.sls.utils.SLSUtils.parseNodesFromSLSTrace(SLSUtils.java:95)
at org.apache.hadoop.yarn.sls.SLSRunner.startNM(SLSRunner.java:181)
at org.apache.hadoop.yarn.sls.SLSRunner.start(SLSRunner.java:139)
at org.apache.hadoop.yarn.sls.SLSRunner.main(SLSRunner.java:524)

 Yarn Scheduler Load Simulator
 -

 Key: YARN-1021
 URL: https://issues.apache.org/jira/browse/YARN-1021
 Project: Hadoop YARN
  Issue Type: New Feature
  Components: scheduler
Reporter: Wei Yan
Assignee: Wei Yan
 Fix For: 2.3.0

 Attachments: YARN-1021-demo.tar.gz, YARN-1021-images.tar.gz, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.pdf


 The Yarn Scheduler is a fertile area of interest with different 
 implementations, e.g., Fifo, Capacity and Fair  schedulers. Meanwhile, 
 several optimizations are also made to improve scheduler performance for 
 different scenarios and workload. Each scheduler algorithm has its own set of 
 features, and drives scheduling decisions by many factors, such as fairness, 
 capacity guarantee, resource availability, etc. It is very important to 
 evaluate a scheduler algorithm very well before we deploy it in a production 
 cluster. Unfortunately, currently it is non-trivial to evaluate a scheduling 
 algorithm. Evaluating in a real cluster is always time and cost consuming, 
 and it is also very hard to find a large-enough cluster. Hence, a simulator 
 which can predict how well a scheduler algorithm for some specific workload 
 would be quite useful.
 We want to build a Scheduler Load Simulator to simulate large-scale Yarn 
 clusters and application loads in a single machine. This would be invaluable 
 in furthering Yarn by providing a tool for researchers and developers to 
 prototype new scheduler features and predict their behavior and performance 
 with reasonable amount of confidence, there-by aiding rapid innovation.
 The simulator will exercise the real Yarn ResourceManager removing the 
 network factor by simulating NodeManagers and ApplicationMasters via handling 
 and dispatching NM/AMs heartbeat events from within the same JVM.
 To keep tracking of scheduler behavior and performance, a scheduler wrapper 
 will wrap the real scheduler.
 The simulator will produce real time metrics while executing, including:
 * Resource usages for whole cluster and each queue, which can be utilized to 
 configure cluster and queue's capacity.
 * The detailed application execution trace (recorded in relation to simulated 
 time), which can be analyzed to understand/validate the  scheduler behavior 
 (individual jobs turn around time, throughput, fairness, capacity guarantee, 
 etc).
 * Several key metrics of scheduler algorithm, such as time cost of each 
 scheduler operation (allocate, handle, etc), which can be utilized by Hadoop 
 developers to find the code spots and scalability limits.
 The simulator will provide real time charts showing the behavior of the 
 scheduler and its performance.
 A short demo is available http://www.youtube.com/watch?v=6thLi8q0qLE, showing 
 how to use simulator to simulate Fair Scheduler and Capacity Scheduler.



--
This message was sent by Atlassian JIRA
(v6.2#6252)


[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2014-03-02 Thread Qi Zhang (JIRA)

[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13917585#comment-13917585
 ] 

Qi Zhang commented on YARN-1021:


Hi @Wei Yan. I am trying to use SLS but always meet with the following 
exception. Can you tell me what is the reason? Thank you!

-bash-3.2$ sudo sh share/hadoop/tools/sls/bin/slsrun.sh 
--input-rumen=share/hadoop/tools/sls/sample-data/2jobs2min-rumen-jh.json 
--output-dir=share/hadoop/tools/sls/sample_output
log4j:WARN No appenders could be found for logger 
(org.apache.hadoop.util.Shell).
log4j:WARN Please initialize the log4j system properly.
log4j:WARN See http://logging.apache.org/log4j/1.2/faq.html#noconfig for more 
info.
Java HotSpot(TM) 64-Bit Server VM warning: You have loaded library 
/usr/local/hadoop-2.3.0/lib/native/libhadoop.so.1.0.0 which might have disabled 
stack guard. The VM will try to fix the stack guard now.
It's highly recommended that you fix the library with 'execstack -c libfile', 
or link it with '-z noexecstack'.
java.lang.NullPointerException
at org.apache.hadoop.yarn.sls.web.SLSWebApp.init(SLSWebApp.java:82)
at 
org.apache.hadoop.yarn.sls.scheduler.ResourceSchedulerWrapper.initMetrics(ResourceSchedulerWrapper.java:463)
at 
org.apache.hadoop.yarn.sls.scheduler.ResourceSchedulerWrapper.setConf(ResourceSchedulerWrapper.java:162)
at 
org.apache.hadoop.util.ReflectionUtils.setConf(ReflectionUtils.java:73)
at 
org.apache.hadoop.util.ReflectionUtils.newInstance(ReflectionUtils.java:133)
at 
org.apache.hadoop.yarn.server.resourcemanager.ResourceManager.createScheduler(ResourceManager.java:230)
at 
org.apache.hadoop.yarn.server.resourcemanager.ResourceManager$RMActiveServices.serviceInit(ResourceManager.java:355)
at 
org.apache.hadoop.service.AbstractService.init(AbstractService.java:163)
at 
org.apache.hadoop.yarn.server.resourcemanager.ResourceManager.createAndInitActiveServices(ResourceManager.java:775)
at 
org.apache.hadoop.yarn.server.resourcemanager.ResourceManager.serviceInit(ResourceManager.java:197)
at 
org.apache.hadoop.service.AbstractService.init(AbstractService.java:163)
at org.apache.hadoop.yarn.sls.SLSRunner.startRM(SLSRunner.java:163)
at org.apache.hadoop.yarn.sls.SLSRunner.start(SLSRunner.java:137)
at org.apache.hadoop.yarn.sls.SLSRunner.main(SLSRunner.java:524)
Exception in thread pool-2-thread-72 java.lang.NullPointerException
at 
org.apache.hadoop.yarn.sls.scheduler.ResourceSchedulerWrapper.addAMRuntime(ResourceSchedulerWrapper.java:721)
at 
org.apache.hadoop.yarn.sls.appmaster.AMSimulator.lastStep(AMSimulator.java:196)
at 
org.apache.hadoop.yarn.sls.appmaster.MRAMSimulator.lastStep(MRAMSimulator.java:390)
at 
org.apache.hadoop.yarn.sls.scheduler.TaskRunner$Task.run(TaskRunner.java:94)
at 
java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1145)
at 
java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:615)
at java.lang.Thread.run(Thread.java:744)
Exception in thread pool-2-thread-98 java.lang.NullPointerException
at 
org.apache.hadoop.yarn.sls.scheduler.ResourceSchedulerWrapper.addAMRuntime(ResourceSchedulerWrapper.java:721)
at 
org.apache.hadoop.yarn.sls.appmaster.AMSimulator.lastStep(AMSimulator.java:196)
at 
org.apache.hadoop.yarn.sls.appmaster.MRAMSimulator.lastStep(MRAMSimulator.java:390)
at 
org.apache.hadoop.yarn.sls.scheduler.TaskRunner$Task.run(TaskRunner.java:94)
at 
java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1145)
at 
java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:615)
at java.lang.Thread.run(Thread.java:744)

 Yarn Scheduler Load Simulator
 -

 Key: YARN-1021
 URL: https://issues.apache.org/jira/browse/YARN-1021
 Project: Hadoop YARN
  Issue Type: New Feature
  Components: scheduler
Reporter: Wei Yan
Assignee: Wei Yan
 Fix For: 2.3.0

 Attachments: YARN-1021-demo.tar.gz, YARN-1021-images.tar.gz, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.pdf


 The Yarn Scheduler is a fertile area of interest with different 
 implementations, e.g., Fifo, Capacity and Fair  schedulers. Meanwhile, 
 several optimizations are also made to improve scheduler performance for 
 different scenarios and workload. Each scheduler algorithm has its own set of 
 features, and drives scheduling decisions by many factors, such as fairness, 
 capacity guarantee, resource 

[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2014-03-02 Thread Wei Yan (JIRA)

[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13917634#comment-13917634
 ] 

Wei Yan commented on YARN-1021:
---

[~qzhang90]. Check the resource simulate.info.html.template. It look the sls 
cannot find it.
And step into the sls directory and try again. cd share/hadoop/tools/sls; 
bin/slsrun.sh --input-rumen=sample-data/2jobs2min-rumen-jh.json 
--output-dir=sample_output.


 Yarn Scheduler Load Simulator
 -

 Key: YARN-1021
 URL: https://issues.apache.org/jira/browse/YARN-1021
 Project: Hadoop YARN
  Issue Type: New Feature
  Components: scheduler
Reporter: Wei Yan
Assignee: Wei Yan
 Fix For: 2.3.0

 Attachments: YARN-1021-demo.tar.gz, YARN-1021-images.tar.gz, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.pdf


 The Yarn Scheduler is a fertile area of interest with different 
 implementations, e.g., Fifo, Capacity and Fair  schedulers. Meanwhile, 
 several optimizations are also made to improve scheduler performance for 
 different scenarios and workload. Each scheduler algorithm has its own set of 
 features, and drives scheduling decisions by many factors, such as fairness, 
 capacity guarantee, resource availability, etc. It is very important to 
 evaluate a scheduler algorithm very well before we deploy it in a production 
 cluster. Unfortunately, currently it is non-trivial to evaluate a scheduling 
 algorithm. Evaluating in a real cluster is always time and cost consuming, 
 and it is also very hard to find a large-enough cluster. Hence, a simulator 
 which can predict how well a scheduler algorithm for some specific workload 
 would be quite useful.
 We want to build a Scheduler Load Simulator to simulate large-scale Yarn 
 clusters and application loads in a single machine. This would be invaluable 
 in furthering Yarn by providing a tool for researchers and developers to 
 prototype new scheduler features and predict their behavior and performance 
 with reasonable amount of confidence, there-by aiding rapid innovation.
 The simulator will exercise the real Yarn ResourceManager removing the 
 network factor by simulating NodeManagers and ApplicationMasters via handling 
 and dispatching NM/AMs heartbeat events from within the same JVM.
 To keep tracking of scheduler behavior and performance, a scheduler wrapper 
 will wrap the real scheduler.
 The simulator will produce real time metrics while executing, including:
 * Resource usages for whole cluster and each queue, which can be utilized to 
 configure cluster and queue's capacity.
 * The detailed application execution trace (recorded in relation to simulated 
 time), which can be analyzed to understand/validate the  scheduler behavior 
 (individual jobs turn around time, throughput, fairness, capacity guarantee, 
 etc).
 * Several key metrics of scheduler algorithm, such as time cost of each 
 scheduler operation (allocate, handle, etc), which can be utilized by Hadoop 
 developers to find the code spots and scalability limits.
 The simulator will provide real time charts showing the behavior of the 
 scheduler and its performance.
 A short demo is available http://www.youtube.com/watch?v=6thLi8q0qLE, showing 
 how to use simulator to simulate Fair Scheduler and Capacity Scheduler.



--
This message was sent by Atlassian JIRA
(v6.2#6252)


[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2014-03-02 Thread Qi Zhang (JIRA)

[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13917641#comment-13917641
 ] 

Qi Zhang commented on YARN-1021:


Wei Yan. Thank you for your suggestion, it solves the problem! 
Actually, I tried to run the slsrun.sh from many other directories expect 
share/hadoop/tools/sls. I think it can be more straightforward if slsrun.sh can 
be executed from any path.

 Yarn Scheduler Load Simulator
 -

 Key: YARN-1021
 URL: https://issues.apache.org/jira/browse/YARN-1021
 Project: Hadoop YARN
  Issue Type: New Feature
  Components: scheduler
Reporter: Wei Yan
Assignee: Wei Yan
 Fix For: 2.3.0

 Attachments: YARN-1021-demo.tar.gz, YARN-1021-images.tar.gz, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.pdf


 The Yarn Scheduler is a fertile area of interest with different 
 implementations, e.g., Fifo, Capacity and Fair  schedulers. Meanwhile, 
 several optimizations are also made to improve scheduler performance for 
 different scenarios and workload. Each scheduler algorithm has its own set of 
 features, and drives scheduling decisions by many factors, such as fairness, 
 capacity guarantee, resource availability, etc. It is very important to 
 evaluate a scheduler algorithm very well before we deploy it in a production 
 cluster. Unfortunately, currently it is non-trivial to evaluate a scheduling 
 algorithm. Evaluating in a real cluster is always time and cost consuming, 
 and it is also very hard to find a large-enough cluster. Hence, a simulator 
 which can predict how well a scheduler algorithm for some specific workload 
 would be quite useful.
 We want to build a Scheduler Load Simulator to simulate large-scale Yarn 
 clusters and application loads in a single machine. This would be invaluable 
 in furthering Yarn by providing a tool for researchers and developers to 
 prototype new scheduler features and predict their behavior and performance 
 with reasonable amount of confidence, there-by aiding rapid innovation.
 The simulator will exercise the real Yarn ResourceManager removing the 
 network factor by simulating NodeManagers and ApplicationMasters via handling 
 and dispatching NM/AMs heartbeat events from within the same JVM.
 To keep tracking of scheduler behavior and performance, a scheduler wrapper 
 will wrap the real scheduler.
 The simulator will produce real time metrics while executing, including:
 * Resource usages for whole cluster and each queue, which can be utilized to 
 configure cluster and queue's capacity.
 * The detailed application execution trace (recorded in relation to simulated 
 time), which can be analyzed to understand/validate the  scheduler behavior 
 (individual jobs turn around time, throughput, fairness, capacity guarantee, 
 etc).
 * Several key metrics of scheduler algorithm, such as time cost of each 
 scheduler operation (allocate, handle, etc), which can be utilized by Hadoop 
 developers to find the code spots and scalability limits.
 The simulator will provide real time charts showing the behavior of the 
 scheduler and its performance.
 A short demo is available http://www.youtube.com/watch?v=6thLi8q0qLE, showing 
 how to use simulator to simulate Fair Scheduler and Capacity Scheduler.



--
This message was sent by Atlassian JIRA
(v6.2#6252)


[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2013-11-06 Thread Carlo Curino (JIRA)

[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13815573#comment-13815573
 ] 

Carlo Curino commented on YARN-1021:


Hi Wei, it would be nice to add in the README how to use the simulator (e.g., 
by having a super simple rumen trace, and configs one can use right away, 
including pointers to the nice visualizations you have). I was looking at it 
today, and while I am sure I can figure it out digging around more, quick 
instructions 
will make it more likely that people pick it up.

 Yarn Scheduler Load Simulator
 -

 Key: YARN-1021
 URL: https://issues.apache.org/jira/browse/YARN-1021
 Project: Hadoop YARN
  Issue Type: New Feature
  Components: scheduler
Reporter: Wei Yan
Assignee: Wei Yan
 Fix For: 2.3.0

 Attachments: YARN-1021-demo.tar.gz, YARN-1021-images.tar.gz, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.pdf


 The Yarn Scheduler is a fertile area of interest with different 
 implementations, e.g., Fifo, Capacity and Fair  schedulers. Meanwhile, 
 several optimizations are also made to improve scheduler performance for 
 different scenarios and workload. Each scheduler algorithm has its own set of 
 features, and drives scheduling decisions by many factors, such as fairness, 
 capacity guarantee, resource availability, etc. It is very important to 
 evaluate a scheduler algorithm very well before we deploy it in a production 
 cluster. Unfortunately, currently it is non-trivial to evaluate a scheduling 
 algorithm. Evaluating in a real cluster is always time and cost consuming, 
 and it is also very hard to find a large-enough cluster. Hence, a simulator 
 which can predict how well a scheduler algorithm for some specific workload 
 would be quite useful.
 We want to build a Scheduler Load Simulator to simulate large-scale Yarn 
 clusters and application loads in a single machine. This would be invaluable 
 in furthering Yarn by providing a tool for researchers and developers to 
 prototype new scheduler features and predict their behavior and performance 
 with reasonable amount of confidence, there-by aiding rapid innovation.
 The simulator will exercise the real Yarn ResourceManager removing the 
 network factor by simulating NodeManagers and ApplicationMasters via handling 
 and dispatching NM/AMs heartbeat events from within the same JVM.
 To keep tracking of scheduler behavior and performance, a scheduler wrapper 
 will wrap the real scheduler.
 The simulator will produce real time metrics while executing, including:
 * Resource usages for whole cluster and each queue, which can be utilized to 
 configure cluster and queue's capacity.
 * The detailed application execution trace (recorded in relation to simulated 
 time), which can be analyzed to understand/validate the  scheduler behavior 
 (individual jobs turn around time, throughput, fairness, capacity guarantee, 
 etc).
 * Several key metrics of scheduler algorithm, such as time cost of each 
 scheduler operation (allocate, handle, etc), which can be utilized by Hadoop 
 developers to find the code spots and scalability limits.
 The simulator will provide real time charts showing the behavior of the 
 scheduler and its performance.
 A short demo is available http://www.youtube.com/watch?v=6thLi8q0qLE, showing 
 how to use simulator to simulate Fair Scheduler and Capacity Scheduler.



--
This message was sent by Atlassian JIRA
(v6.1#6144)


[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2013-11-06 Thread Wei Yan (JIRA)

[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13815592#comment-13815592
 ] 

Wei Yan commented on YARN-1021:
---

[~curino], thanks for your comment.

Currently we put the instructions in the .pdf document and the generated site 
document. We already have a simple rumen trace 
(hadoop-sls/src/main/data/2job2min-rumen-jh.json) and example configurations 
(hadoop-sls/src/main/sample-conf). I'll make this clear in the README.

YARN-1393 is created for this issue.


 Yarn Scheduler Load Simulator
 -

 Key: YARN-1021
 URL: https://issues.apache.org/jira/browse/YARN-1021
 Project: Hadoop YARN
  Issue Type: New Feature
  Components: scheduler
Reporter: Wei Yan
Assignee: Wei Yan
 Fix For: 2.3.0

 Attachments: YARN-1021-demo.tar.gz, YARN-1021-images.tar.gz, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.pdf


 The Yarn Scheduler is a fertile area of interest with different 
 implementations, e.g., Fifo, Capacity and Fair  schedulers. Meanwhile, 
 several optimizations are also made to improve scheduler performance for 
 different scenarios and workload. Each scheduler algorithm has its own set of 
 features, and drives scheduling decisions by many factors, such as fairness, 
 capacity guarantee, resource availability, etc. It is very important to 
 evaluate a scheduler algorithm very well before we deploy it in a production 
 cluster. Unfortunately, currently it is non-trivial to evaluate a scheduling 
 algorithm. Evaluating in a real cluster is always time and cost consuming, 
 and it is also very hard to find a large-enough cluster. Hence, a simulator 
 which can predict how well a scheduler algorithm for some specific workload 
 would be quite useful.
 We want to build a Scheduler Load Simulator to simulate large-scale Yarn 
 clusters and application loads in a single machine. This would be invaluable 
 in furthering Yarn by providing a tool for researchers and developers to 
 prototype new scheduler features and predict their behavior and performance 
 with reasonable amount of confidence, there-by aiding rapid innovation.
 The simulator will exercise the real Yarn ResourceManager removing the 
 network factor by simulating NodeManagers and ApplicationMasters via handling 
 and dispatching NM/AMs heartbeat events from within the same JVM.
 To keep tracking of scheduler behavior and performance, a scheduler wrapper 
 will wrap the real scheduler.
 The simulator will produce real time metrics while executing, including:
 * Resource usages for whole cluster and each queue, which can be utilized to 
 configure cluster and queue's capacity.
 * The detailed application execution trace (recorded in relation to simulated 
 time), which can be analyzed to understand/validate the  scheduler behavior 
 (individual jobs turn around time, throughput, fairness, capacity guarantee, 
 etc).
 * Several key metrics of scheduler algorithm, such as time cost of each 
 scheduler operation (allocate, handle, etc), which can be utilized by Hadoop 
 developers to find the code spots and scalability limits.
 The simulator will provide real time charts showing the behavior of the 
 scheduler and its performance.
 A short demo is available http://www.youtube.com/watch?v=6thLi8q0qLE, showing 
 how to use simulator to simulate Fair Scheduler and Capacity Scheduler.



--
This message was sent by Atlassian JIRA
(v6.1#6144)


[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2013-10-04 Thread Wei Yan (JIRA)

[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13786407#comment-13786407
 ] 

Wei Yan commented on YARN-1021:
---

Thanks, [~mitdesai].
I'll look into it.

 Yarn Scheduler Load Simulator
 -

 Key: YARN-1021
 URL: https://issues.apache.org/jira/browse/YARN-1021
 Project: Hadoop YARN
  Issue Type: New Feature
  Components: scheduler
Reporter: Wei Yan
Assignee: Wei Yan
 Fix For: 2.3.0

 Attachments: YARN-1021-demo.tar.gz, YARN-1021-images.tar.gz, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.pdf


 The Yarn Scheduler is a fertile area of interest with different 
 implementations, e.g., Fifo, Capacity and Fair  schedulers. Meanwhile, 
 several optimizations are also made to improve scheduler performance for 
 different scenarios and workload. Each scheduler algorithm has its own set of 
 features, and drives scheduling decisions by many factors, such as fairness, 
 capacity guarantee, resource availability, etc. It is very important to 
 evaluate a scheduler algorithm very well before we deploy it in a production 
 cluster. Unfortunately, currently it is non-trivial to evaluate a scheduling 
 algorithm. Evaluating in a real cluster is always time and cost consuming, 
 and it is also very hard to find a large-enough cluster. Hence, a simulator 
 which can predict how well a scheduler algorithm for some specific workload 
 would be quite useful.
 We want to build a Scheduler Load Simulator to simulate large-scale Yarn 
 clusters and application loads in a single machine. This would be invaluable 
 in furthering Yarn by providing a tool for researchers and developers to 
 prototype new scheduler features and predict their behavior and performance 
 with reasonable amount of confidence, there-by aiding rapid innovation.
 The simulator will exercise the real Yarn ResourceManager removing the 
 network factor by simulating NodeManagers and ApplicationMasters via handling 
 and dispatching NM/AMs heartbeat events from within the same JVM.
 To keep tracking of scheduler behavior and performance, a scheduler wrapper 
 will wrap the real scheduler.
 The simulator will produce real time metrics while executing, including:
 * Resource usages for whole cluster and each queue, which can be utilized to 
 configure cluster and queue's capacity.
 * The detailed application execution trace (recorded in relation to simulated 
 time), which can be analyzed to understand/validate the  scheduler behavior 
 (individual jobs turn around time, throughput, fairness, capacity guarantee, 
 etc).
 * Several key metrics of scheduler algorithm, such as time cost of each 
 scheduler operation (allocate, handle, etc), which can be utilized by Hadoop 
 developers to find the code spots and scalability limits.
 The simulator will provide real time charts showing the behavior of the 
 scheduler and its performance.
 A short demo is available http://www.youtube.com/watch?v=6thLi8q0qLE, showing 
 how to use simulator to simulate Fair Scheduler and Capacity Scheduler.



--
This message was sent by Atlassian JIRA
(v6.1#6144)


[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2013-10-03 Thread Mit Desai (JIRA)

[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13785547#comment-13785547
 ] 

Mit Desai commented on YARN-1021:
-

Hey Wei, FYI, I would like to inform you that the test TestSLSRunner is 
failing. I have created a new JIRA for that YARN-1270

 Yarn Scheduler Load Simulator
 -

 Key: YARN-1021
 URL: https://issues.apache.org/jira/browse/YARN-1021
 Project: Hadoop YARN
  Issue Type: New Feature
  Components: scheduler
Reporter: Wei Yan
Assignee: Wei Yan
 Fix For: 2.3.0

 Attachments: YARN-1021-demo.tar.gz, YARN-1021-images.tar.gz, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.pdf


 The Yarn Scheduler is a fertile area of interest with different 
 implementations, e.g., Fifo, Capacity and Fair  schedulers. Meanwhile, 
 several optimizations are also made to improve scheduler performance for 
 different scenarios and workload. Each scheduler algorithm has its own set of 
 features, and drives scheduling decisions by many factors, such as fairness, 
 capacity guarantee, resource availability, etc. It is very important to 
 evaluate a scheduler algorithm very well before we deploy it in a production 
 cluster. Unfortunately, currently it is non-trivial to evaluate a scheduling 
 algorithm. Evaluating in a real cluster is always time and cost consuming, 
 and it is also very hard to find a large-enough cluster. Hence, a simulator 
 which can predict how well a scheduler algorithm for some specific workload 
 would be quite useful.
 We want to build a Scheduler Load Simulator to simulate large-scale Yarn 
 clusters and application loads in a single machine. This would be invaluable 
 in furthering Yarn by providing a tool for researchers and developers to 
 prototype new scheduler features and predict their behavior and performance 
 with reasonable amount of confidence, there-by aiding rapid innovation.
 The simulator will exercise the real Yarn ResourceManager removing the 
 network factor by simulating NodeManagers and ApplicationMasters via handling 
 and dispatching NM/AMs heartbeat events from within the same JVM.
 To keep tracking of scheduler behavior and performance, a scheduler wrapper 
 will wrap the real scheduler.
 The simulator will produce real time metrics while executing, including:
 * Resource usages for whole cluster and each queue, which can be utilized to 
 configure cluster and queue's capacity.
 * The detailed application execution trace (recorded in relation to simulated 
 time), which can be analyzed to understand/validate the  scheduler behavior 
 (individual jobs turn around time, throughput, fairness, capacity guarantee, 
 etc).
 * Several key metrics of scheduler algorithm, such as time cost of each 
 scheduler operation (allocate, handle, etc), which can be utilized by Hadoop 
 developers to find the code spots and scalability limits.
 The simulator will provide real time charts showing the behavior of the 
 scheduler and its performance.
 A short demo is available http://www.youtube.com/watch?v=6thLi8q0qLE, showing 
 how to use simulator to simulate Fair Scheduler and Capacity Scheduler.



--
This message was sent by Atlassian JIRA
(v6.1#6144)


[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2013-09-28 Thread Hudson (JIRA)

[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13780781#comment-13780781
 ] 

Hudson commented on YARN-1021:
--

FAILURE: Integrated in Hadoop-Yarn-trunk #346 (See 
[https://builds.apache.org/job/Hadoop-Yarn-trunk/346/])
YARN-1021. Yarn Scheduler Load Simulator. (ywskycn via tucu) (tucu: 
http://svn.apache.org/viewcvs.cgi/?root=Apache-SVNview=revrev=1527059)
* 
/hadoop/common/trunk/hadoop-assemblies/src/main/resources/assemblies/hadoop-sls.xml
* 
/hadoop/common/trunk/hadoop-assemblies/src/main/resources/assemblies/hadoop-tools.xml
* /hadoop/common/trunk/hadoop-project/pom.xml
* /hadoop/common/trunk/hadoop-project/src/site/site.xml
* /hadoop/common/trunk/hadoop-tools/hadoop-openstack
* /hadoop/common/trunk/hadoop-tools/hadoop-sls
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/README
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/dev-support
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/dev-support/findbugs-exclude.xml
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/pom.xml
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/assemblies
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/assemblies/sls.xml
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/bin
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/bin/rumen2sls.sh
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/bin/slsrun.sh
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/data
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/data/2jobs2min-rumen-jh.json
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html/css
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html/css/bootstrap-responsive.min.css
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html/css/bootstrap.min.css
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html/js
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html/js/thirdparty
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html/js/thirdparty/bootstrap.min.js
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html/js/thirdparty/d3-LICENSE
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html/js/thirdparty/d3.v3.js
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html/js/thirdparty/jquery.js
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html/showSimulationTrace.html
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html/simulate.html.template
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html/simulate.info.html.template
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html/track.html.template
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/RumenToSLSConverter.java
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/SLSRunner.java
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/appmaster
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/appmaster/AMSimulator.java
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/appmaster/MRAMSimulator.java
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/conf
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/conf/SLSConfiguration.java
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/nodemanager
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/nodemanager/NMSimulator.java
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/nodemanager/NodeInfo.java
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/scheduler
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/scheduler/CapacitySchedulerMetrics.java
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/scheduler/ContainerSimulator.java
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/scheduler/FairSchedulerMetrics.java
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/scheduler/FifoSchedulerMetrics.java
* 

[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2013-09-28 Thread Hudson (JIRA)

[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13780836#comment-13780836
 ] 

Hudson commented on YARN-1021:
--

SUCCESS: Integrated in Hadoop-Hdfs-trunk #1536 (See 
[https://builds.apache.org/job/Hadoop-Hdfs-trunk/1536/])
YARN-1021. Yarn Scheduler Load Simulator. (ywskycn via tucu) (tucu: 
http://svn.apache.org/viewcvs.cgi/?root=Apache-SVNview=revrev=1527059)
* 
/hadoop/common/trunk/hadoop-assemblies/src/main/resources/assemblies/hadoop-sls.xml
* 
/hadoop/common/trunk/hadoop-assemblies/src/main/resources/assemblies/hadoop-tools.xml
* /hadoop/common/trunk/hadoop-project/pom.xml
* /hadoop/common/trunk/hadoop-project/src/site/site.xml
* /hadoop/common/trunk/hadoop-tools/hadoop-openstack
* /hadoop/common/trunk/hadoop-tools/hadoop-sls
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/README
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/dev-support
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/dev-support/findbugs-exclude.xml
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/pom.xml
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/assemblies
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/assemblies/sls.xml
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/bin
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/bin/rumen2sls.sh
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/bin/slsrun.sh
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/data
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/data/2jobs2min-rumen-jh.json
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html/css
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html/css/bootstrap-responsive.min.css
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html/css/bootstrap.min.css
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html/js
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html/js/thirdparty
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html/js/thirdparty/bootstrap.min.js
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html/js/thirdparty/d3-LICENSE
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html/js/thirdparty/d3.v3.js
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html/js/thirdparty/jquery.js
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html/showSimulationTrace.html
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html/simulate.html.template
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html/simulate.info.html.template
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html/track.html.template
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/RumenToSLSConverter.java
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/SLSRunner.java
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/appmaster
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/appmaster/AMSimulator.java
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/appmaster/MRAMSimulator.java
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/conf
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/conf/SLSConfiguration.java
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/nodemanager
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/nodemanager/NMSimulator.java
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/nodemanager/NodeInfo.java
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/scheduler
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/scheduler/CapacitySchedulerMetrics.java
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/scheduler/ContainerSimulator.java
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/scheduler/FairSchedulerMetrics.java
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/scheduler/FifoSchedulerMetrics.java
* 

[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2013-09-28 Thread Hudson (JIRA)

[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13780848#comment-13780848
 ] 

Hudson commented on YARN-1021:
--

FAILURE: Integrated in Hadoop-Mapreduce-trunk #1562 (See 
[https://builds.apache.org/job/Hadoop-Mapreduce-trunk/1562/])
YARN-1021. Yarn Scheduler Load Simulator. (ywskycn via tucu) (tucu: 
http://svn.apache.org/viewcvs.cgi/?root=Apache-SVNview=revrev=1527059)
* 
/hadoop/common/trunk/hadoop-assemblies/src/main/resources/assemblies/hadoop-sls.xml
* 
/hadoop/common/trunk/hadoop-assemblies/src/main/resources/assemblies/hadoop-tools.xml
* /hadoop/common/trunk/hadoop-project/pom.xml
* /hadoop/common/trunk/hadoop-project/src/site/site.xml
* /hadoop/common/trunk/hadoop-tools/hadoop-openstack
* /hadoop/common/trunk/hadoop-tools/hadoop-sls
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/README
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/dev-support
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/dev-support/findbugs-exclude.xml
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/pom.xml
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/assemblies
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/assemblies/sls.xml
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/bin
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/bin/rumen2sls.sh
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/bin/slsrun.sh
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/data
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/data/2jobs2min-rumen-jh.json
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html/css
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html/css/bootstrap-responsive.min.css
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html/css/bootstrap.min.css
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html/js
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html/js/thirdparty
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html/js/thirdparty/bootstrap.min.js
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html/js/thirdparty/d3-LICENSE
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html/js/thirdparty/d3.v3.js
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html/js/thirdparty/jquery.js
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html/showSimulationTrace.html
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html/simulate.html.template
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html/simulate.info.html.template
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html/track.html.template
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/RumenToSLSConverter.java
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/SLSRunner.java
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/appmaster
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/appmaster/AMSimulator.java
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/appmaster/MRAMSimulator.java
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/conf
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/conf/SLSConfiguration.java
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/nodemanager
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/nodemanager/NMSimulator.java
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/nodemanager/NodeInfo.java
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/scheduler
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/scheduler/CapacitySchedulerMetrics.java
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/scheduler/ContainerSimulator.java
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/scheduler/FairSchedulerMetrics.java
* 

[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2013-09-27 Thread Hudson (JIRA)

[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13780347#comment-13780347
 ] 

Hudson commented on YARN-1021:
--

SUCCESS: Integrated in Hadoop-trunk-Commit #4487 (See 
[https://builds.apache.org/job/Hadoop-trunk-Commit/4487/])
YARN-1021. Yarn Scheduler Load Simulator. (ywskycn via tucu) (tucu: 
http://svn.apache.org/viewcvs.cgi/?root=Apache-SVNview=revrev=1527059)
* 
/hadoop/common/trunk/hadoop-assemblies/src/main/resources/assemblies/hadoop-sls.xml
* 
/hadoop/common/trunk/hadoop-assemblies/src/main/resources/assemblies/hadoop-tools.xml
* /hadoop/common/trunk/hadoop-project/pom.xml
* /hadoop/common/trunk/hadoop-project/src/site/site.xml
* /hadoop/common/trunk/hadoop-tools/hadoop-openstack
* /hadoop/common/trunk/hadoop-tools/hadoop-sls
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/README
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/dev-support
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/dev-support/findbugs-exclude.xml
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/pom.xml
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/assemblies
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/assemblies/sls.xml
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/bin
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/bin/rumen2sls.sh
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/bin/slsrun.sh
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/data
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/data/2jobs2min-rumen-jh.json
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html/css
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html/css/bootstrap-responsive.min.css
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html/css/bootstrap.min.css
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html/js
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html/js/thirdparty
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html/js/thirdparty/bootstrap.min.js
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html/js/thirdparty/d3-LICENSE
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html/js/thirdparty/d3.v3.js
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html/js/thirdparty/jquery.js
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html/showSimulationTrace.html
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html/simulate.html.template
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html/simulate.info.html.template
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/html/track.html.template
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache
* /hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/RumenToSLSConverter.java
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/SLSRunner.java
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/appmaster
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/appmaster/AMSimulator.java
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/appmaster/MRAMSimulator.java
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/conf
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/conf/SLSConfiguration.java
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/nodemanager
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/nodemanager/NMSimulator.java
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/nodemanager/NodeInfo.java
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/scheduler
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/scheduler/CapacitySchedulerMetrics.java
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/scheduler/ContainerSimulator.java
* 
/hadoop/common/trunk/hadoop-tools/hadoop-sls/src/main/java/org/apache/hadoop/yarn/sls/scheduler/FairSchedulerMetrics.java
* 

[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2013-09-24 Thread Hadoop QA (JIRA)

[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13776142#comment-13776142
 ] 

Hadoop QA commented on YARN-1021:
-

{color:red}-1 overall{color}.  Here are the results of testing the latest 
attachment 
  http://issues.apache.org/jira/secure/attachment/12604747/YARN-1021.patch
  against trunk revision .

{color:green}+1 @author{color}.  The patch does not contain any @author 
tags.

{color:green}+1 tests included{color}.  The patch appears to include 11 new 
or modified test files.

  {color:red}-1 javac{color}.  The applied patch generated 1149 javac 
compiler warnings (more than the trunk's current 1145 warnings).

{color:green}+1 javadoc{color}.  The javadoc tool did not generate any 
warning messages.

{color:green}+1 eclipse:eclipse{color}.  The patch built with 
eclipse:eclipse.

{color:green}+1 findbugs{color}.  The patch does not introduce any new 
Findbugs (version 1.3.9) warnings.

{color:green}+1 release audit{color}.  The applied patch does not increase 
the total number of release audit warnings.

{color:green}+1 core tests{color}.  The patch passed unit tests in 
hadoop-assemblies hadoop-tools/hadoop-sls hadoop-tools/hadoop-tools-dist.

{color:green}+1 contrib tests{color}.  The patch passed contrib unit tests.

Test results: 
https://builds.apache.org/job/PreCommit-YARN-Build/1995//testReport/
Javac warnings: 
https://builds.apache.org/job/PreCommit-YARN-Build/1995//artifact/trunk/patchprocess/diffJavacWarnings.txt
Console output: https://builds.apache.org/job/PreCommit-YARN-Build/1995//console

This message is automatically generated.

 Yarn Scheduler Load Simulator
 -

 Key: YARN-1021
 URL: https://issues.apache.org/jira/browse/YARN-1021
 Project: Hadoop YARN
  Issue Type: New Feature
  Components: scheduler
Reporter: Wei Yan
Assignee: Wei Yan
 Attachments: YARN-1021-demo.tar.gz, YARN-1021-images.tar.gz, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.pdf


 The Yarn Scheduler is a fertile area of interest with different 
 implementations, e.g., Fifo, Capacity and Fair  schedulers. Meanwhile, 
 several optimizations are also made to improve scheduler performance for 
 different scenarios and workload. Each scheduler algorithm has its own set of 
 features, and drives scheduling decisions by many factors, such as fairness, 
 capacity guarantee, resource availability, etc. It is very important to 
 evaluate a scheduler algorithm very well before we deploy it in a production 
 cluster. Unfortunately, currently it is non-trivial to evaluate a scheduling 
 algorithm. Evaluating in a real cluster is always time and cost consuming, 
 and it is also very hard to find a large-enough cluster. Hence, a simulator 
 which can predict how well a scheduler algorithm for some specific workload 
 would be quite useful.
 We want to build a Scheduler Load Simulator to simulate large-scale Yarn 
 clusters and application loads in a single machine. This would be invaluable 
 in furthering Yarn by providing a tool for researchers and developers to 
 prototype new scheduler features and predict their behavior and performance 
 with reasonable amount of confidence, there-by aiding rapid innovation.
 The simulator will exercise the real Yarn ResourceManager removing the 
 network factor by simulating NodeManagers and ApplicationMasters via handling 
 and dispatching NM/AMs heartbeat events from within the same JVM.
 To keep tracking of scheduler behavior and performance, a scheduler wrapper 
 will wrap the real scheduler.
 The simulator will produce real time metrics while executing, including:
 * Resource usages for whole cluster and each queue, which can be utilized to 
 configure cluster and queue's capacity.
 * The detailed application execution trace (recorded in relation to simulated 
 time), which can be analyzed to understand/validate the  scheduler behavior 
 (individual jobs turn around time, throughput, fairness, capacity guarantee, 
 etc).
 * Several key metrics of scheduler algorithm, such as time cost of each 
 scheduler operation (allocate, handle, etc), which can be utilized by Hadoop 
 developers to find the code spots and scalability limits.
 The simulator will provide real time charts showing the behavior of the 
 scheduler and its performance.
 A short demo is available http://www.youtube.com/watch?v=6thLi8q0qLE, showing 
 how to use simulator to simulate Fair Scheduler and Capacity Scheduler.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more 

[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2013-09-24 Thread Hadoop QA (JIRA)

[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13776173#comment-13776173
 ] 

Hadoop QA commented on YARN-1021:
-

{color:red}-1 overall{color}.  Here are the results of testing the latest 
attachment 
  http://issues.apache.org/jira/secure/attachment/12604767/YARN-1021.patch
  against trunk revision .

{color:green}+1 @author{color}.  The patch does not contain any @author 
tags.

{color:green}+1 tests included{color}.  The patch appears to include 11 new 
or modified test files.

{color:green}+1 javac{color}.  The applied patch does not increase the 
total number of javac compiler warnings.

{color:green}+1 javadoc{color}.  The javadoc tool did not generate any 
warning messages.

{color:green}+1 eclipse:eclipse{color}.  The patch built with 
eclipse:eclipse.

{color:green}+1 findbugs{color}.  The patch does not introduce any new 
Findbugs (version 1.3.9) warnings.

{color:green}+1 release audit{color}.  The applied patch does not increase 
the total number of release audit warnings.

{color:red}-1 core tests{color}.  The patch failed these unit tests in 
hadoop-assemblies hadoop-tools/hadoop-sls hadoop-tools/hadoop-tools-dist:

  org.apache.hadoop.yarn.sls.TestSLSRunner

{color:green}+1 contrib tests{color}.  The patch passed contrib unit tests.

Test results: 
https://builds.apache.org/job/PreCommit-YARN-Build/1996//testReport/
Console output: https://builds.apache.org/job/PreCommit-YARN-Build/1996//console

This message is automatically generated.

 Yarn Scheduler Load Simulator
 -

 Key: YARN-1021
 URL: https://issues.apache.org/jira/browse/YARN-1021
 Project: Hadoop YARN
  Issue Type: New Feature
  Components: scheduler
Reporter: Wei Yan
Assignee: Wei Yan
 Attachments: YARN-1021-demo.tar.gz, YARN-1021-images.tar.gz, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.pdf


 The Yarn Scheduler is a fertile area of interest with different 
 implementations, e.g., Fifo, Capacity and Fair  schedulers. Meanwhile, 
 several optimizations are also made to improve scheduler performance for 
 different scenarios and workload. Each scheduler algorithm has its own set of 
 features, and drives scheduling decisions by many factors, such as fairness, 
 capacity guarantee, resource availability, etc. It is very important to 
 evaluate a scheduler algorithm very well before we deploy it in a production 
 cluster. Unfortunately, currently it is non-trivial to evaluate a scheduling 
 algorithm. Evaluating in a real cluster is always time and cost consuming, 
 and it is also very hard to find a large-enough cluster. Hence, a simulator 
 which can predict how well a scheduler algorithm for some specific workload 
 would be quite useful.
 We want to build a Scheduler Load Simulator to simulate large-scale Yarn 
 clusters and application loads in a single machine. This would be invaluable 
 in furthering Yarn by providing a tool for researchers and developers to 
 prototype new scheduler features and predict their behavior and performance 
 with reasonable amount of confidence, there-by aiding rapid innovation.
 The simulator will exercise the real Yarn ResourceManager removing the 
 network factor by simulating NodeManagers and ApplicationMasters via handling 
 and dispatching NM/AMs heartbeat events from within the same JVM.
 To keep tracking of scheduler behavior and performance, a scheduler wrapper 
 will wrap the real scheduler.
 The simulator will produce real time metrics while executing, including:
 * Resource usages for whole cluster and each queue, which can be utilized to 
 configure cluster and queue's capacity.
 * The detailed application execution trace (recorded in relation to simulated 
 time), which can be analyzed to understand/validate the  scheduler behavior 
 (individual jobs turn around time, throughput, fairness, capacity guarantee, 
 etc).
 * Several key metrics of scheduler algorithm, such as time cost of each 
 scheduler operation (allocate, handle, etc), which can be utilized by Hadoop 
 developers to find the code spots and scalability limits.
 The simulator will provide real time charts showing the behavior of the 
 scheduler and its performance.
 A short demo is available http://www.youtube.com/watch?v=6thLi8q0qLE, showing 
 how to use simulator to simulate Fair Scheduler and Capacity Scheduler.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira


[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2013-09-24 Thread Hadoop QA (JIRA)

[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13776357#comment-13776357
 ] 

Hadoop QA commented on YARN-1021:
-

{color:green}+1 overall{color}.  Here are the results of testing the latest 
attachment 
  http://issues.apache.org/jira/secure/attachment/12604801/YARN-1021.patch
  against trunk revision .

{color:green}+1 @author{color}.  The patch does not contain any @author 
tags.

{color:green}+1 tests included{color}.  The patch appears to include 11 new 
or modified test files.

{color:green}+1 javac{color}.  The applied patch does not increase the 
total number of javac compiler warnings.

{color:green}+1 javadoc{color}.  The javadoc tool did not generate any 
warning messages.

{color:green}+1 eclipse:eclipse{color}.  The patch built with 
eclipse:eclipse.

{color:green}+1 findbugs{color}.  The patch does not introduce any new 
Findbugs (version 1.3.9) warnings.

{color:green}+1 release audit{color}.  The applied patch does not increase 
the total number of release audit warnings.

{color:green}+1 core tests{color}.  The patch passed unit tests in 
hadoop-assemblies hadoop-tools/hadoop-sls hadoop-tools/hadoop-tools-dist.

{color:green}+1 contrib tests{color}.  The patch passed contrib unit tests.

Test results: 
https://builds.apache.org/job/PreCommit-YARN-Build/1998//testReport/
Console output: https://builds.apache.org/job/PreCommit-YARN-Build/1998//console

This message is automatically generated.

 Yarn Scheduler Load Simulator
 -

 Key: YARN-1021
 URL: https://issues.apache.org/jira/browse/YARN-1021
 Project: Hadoop YARN
  Issue Type: New Feature
  Components: scheduler
Reporter: Wei Yan
Assignee: Wei Yan
 Attachments: YARN-1021-demo.tar.gz, YARN-1021-images.tar.gz, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.pdf


 The Yarn Scheduler is a fertile area of interest with different 
 implementations, e.g., Fifo, Capacity and Fair  schedulers. Meanwhile, 
 several optimizations are also made to improve scheduler performance for 
 different scenarios and workload. Each scheduler algorithm has its own set of 
 features, and drives scheduling decisions by many factors, such as fairness, 
 capacity guarantee, resource availability, etc. It is very important to 
 evaluate a scheduler algorithm very well before we deploy it in a production 
 cluster. Unfortunately, currently it is non-trivial to evaluate a scheduling 
 algorithm. Evaluating in a real cluster is always time and cost consuming, 
 and it is also very hard to find a large-enough cluster. Hence, a simulator 
 which can predict how well a scheduler algorithm for some specific workload 
 would be quite useful.
 We want to build a Scheduler Load Simulator to simulate large-scale Yarn 
 clusters and application loads in a single machine. This would be invaluable 
 in furthering Yarn by providing a tool for researchers and developers to 
 prototype new scheduler features and predict their behavior and performance 
 with reasonable amount of confidence, there-by aiding rapid innovation.
 The simulator will exercise the real Yarn ResourceManager removing the 
 network factor by simulating NodeManagers and ApplicationMasters via handling 
 and dispatching NM/AMs heartbeat events from within the same JVM.
 To keep tracking of scheduler behavior and performance, a scheduler wrapper 
 will wrap the real scheduler.
 The simulator will produce real time metrics while executing, including:
 * Resource usages for whole cluster and each queue, which can be utilized to 
 configure cluster and queue's capacity.
 * The detailed application execution trace (recorded in relation to simulated 
 time), which can be analyzed to understand/validate the  scheduler behavior 
 (individual jobs turn around time, throughput, fairness, capacity guarantee, 
 etc).
 * Several key metrics of scheduler algorithm, such as time cost of each 
 scheduler operation (allocate, handle, etc), which can be utilized by Hadoop 
 developers to find the code spots and scalability limits.
 The simulator will provide real time charts showing the behavior of the 
 scheduler and its performance.
 A short demo is available http://www.youtube.com/watch?v=6thLi8q0qLE, showing 
 how to use simulator to simulate Fair Scheduler and Capacity Scheduler.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira


[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2013-09-24 Thread Wei Yan (JIRA)

[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13776364#comment-13776364
 ] 

Wei Yan commented on YARN-1021:
---

Update a new patch according to [~tucu00]'s latest comments.
And also let simulator support two types of inputs:
(1) The rumen traces, thus users can directly deploy their rumen traces to the 
simulator.
(2) The simulator itself traces (sls), which is much simpler and users can 
easily generate various workloads. The simulator also has a tool to help users 
convert rumen traces to sls traces.

 Yarn Scheduler Load Simulator
 -

 Key: YARN-1021
 URL: https://issues.apache.org/jira/browse/YARN-1021
 Project: Hadoop YARN
  Issue Type: New Feature
  Components: scheduler
Reporter: Wei Yan
Assignee: Wei Yan
 Attachments: YARN-1021-demo.tar.gz, YARN-1021-images.tar.gz, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.pdf


 The Yarn Scheduler is a fertile area of interest with different 
 implementations, e.g., Fifo, Capacity and Fair  schedulers. Meanwhile, 
 several optimizations are also made to improve scheduler performance for 
 different scenarios and workload. Each scheduler algorithm has its own set of 
 features, and drives scheduling decisions by many factors, such as fairness, 
 capacity guarantee, resource availability, etc. It is very important to 
 evaluate a scheduler algorithm very well before we deploy it in a production 
 cluster. Unfortunately, currently it is non-trivial to evaluate a scheduling 
 algorithm. Evaluating in a real cluster is always time and cost consuming, 
 and it is also very hard to find a large-enough cluster. Hence, a simulator 
 which can predict how well a scheduler algorithm for some specific workload 
 would be quite useful.
 We want to build a Scheduler Load Simulator to simulate large-scale Yarn 
 clusters and application loads in a single machine. This would be invaluable 
 in furthering Yarn by providing a tool for researchers and developers to 
 prototype new scheduler features and predict their behavior and performance 
 with reasonable amount of confidence, there-by aiding rapid innovation.
 The simulator will exercise the real Yarn ResourceManager removing the 
 network factor by simulating NodeManagers and ApplicationMasters via handling 
 and dispatching NM/AMs heartbeat events from within the same JVM.
 To keep tracking of scheduler behavior and performance, a scheduler wrapper 
 will wrap the real scheduler.
 The simulator will produce real time metrics while executing, including:
 * Resource usages for whole cluster and each queue, which can be utilized to 
 configure cluster and queue's capacity.
 * The detailed application execution trace (recorded in relation to simulated 
 time), which can be analyzed to understand/validate the  scheduler behavior 
 (individual jobs turn around time, throughput, fairness, capacity guarantee, 
 etc).
 * Several key metrics of scheduler algorithm, such as time cost of each 
 scheduler operation (allocate, handle, etc), which can be utilized by Hadoop 
 developers to find the code spots and scalability limits.
 The simulator will provide real time charts showing the behavior of the 
 scheduler and its performance.
 A short demo is available http://www.youtube.com/watch?v=6thLi8q0qLE, showing 
 how to use simulator to simulate Fair Scheduler and Capacity Scheduler.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira


[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2013-09-24 Thread Alejandro Abdelnur (JIRA)

[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13776367#comment-13776367
 ] 

Alejandro Abdelnur commented on YARN-1021:
--

[~ywskycn], we shouldn't use /tmp as that does not get clean up by the build, 
instead we should use a temp subdir under target/, easily done by:

{code}
File dir = new File(target, UUID.randomUUID());
dir.mkdirs();
{code}

And the documentation, in the appendix should have a complete/simple example of 
an sls JSON input file as a reference.

 Yarn Scheduler Load Simulator
 -

 Key: YARN-1021
 URL: https://issues.apache.org/jira/browse/YARN-1021
 Project: Hadoop YARN
  Issue Type: New Feature
  Components: scheduler
Reporter: Wei Yan
Assignee: Wei Yan
 Attachments: YARN-1021-demo.tar.gz, YARN-1021-images.tar.gz, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.pdf


 The Yarn Scheduler is a fertile area of interest with different 
 implementations, e.g., Fifo, Capacity and Fair  schedulers. Meanwhile, 
 several optimizations are also made to improve scheduler performance for 
 different scenarios and workload. Each scheduler algorithm has its own set of 
 features, and drives scheduling decisions by many factors, such as fairness, 
 capacity guarantee, resource availability, etc. It is very important to 
 evaluate a scheduler algorithm very well before we deploy it in a production 
 cluster. Unfortunately, currently it is non-trivial to evaluate a scheduling 
 algorithm. Evaluating in a real cluster is always time and cost consuming, 
 and it is also very hard to find a large-enough cluster. Hence, a simulator 
 which can predict how well a scheduler algorithm for some specific workload 
 would be quite useful.
 We want to build a Scheduler Load Simulator to simulate large-scale Yarn 
 clusters and application loads in a single machine. This would be invaluable 
 in furthering Yarn by providing a tool for researchers and developers to 
 prototype new scheduler features and predict their behavior and performance 
 with reasonable amount of confidence, there-by aiding rapid innovation.
 The simulator will exercise the real Yarn ResourceManager removing the 
 network factor by simulating NodeManagers and ApplicationMasters via handling 
 and dispatching NM/AMs heartbeat events from within the same JVM.
 To keep tracking of scheduler behavior and performance, a scheduler wrapper 
 will wrap the real scheduler.
 The simulator will produce real time metrics while executing, including:
 * Resource usages for whole cluster and each queue, which can be utilized to 
 configure cluster and queue's capacity.
 * The detailed application execution trace (recorded in relation to simulated 
 time), which can be analyzed to understand/validate the  scheduler behavior 
 (individual jobs turn around time, throughput, fairness, capacity guarantee, 
 etc).
 * Several key metrics of scheduler algorithm, such as time cost of each 
 scheduler operation (allocate, handle, etc), which can be utilized by Hadoop 
 developers to find the code spots and scalability limits.
 The simulator will provide real time charts showing the behavior of the 
 scheduler and its performance.
 A short demo is available http://www.youtube.com/watch?v=6thLi8q0qLE, showing 
 how to use simulator to simulate Fair Scheduler and Capacity Scheduler.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira


[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2013-09-24 Thread Hadoop QA (JIRA)

[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13776441#comment-13776441
 ] 

Hadoop QA commented on YARN-1021:
-

{color:green}+1 overall{color}.  Here are the results of testing the latest 
attachment 
  http://issues.apache.org/jira/secure/attachment/12604818/YARN-1021.patch
  against trunk revision .

{color:green}+1 @author{color}.  The patch does not contain any @author 
tags.

{color:green}+1 tests included{color}.  The patch appears to include 11 new 
or modified test files.

{color:green}+1 javac{color}.  The applied patch does not increase the 
total number of javac compiler warnings.

{color:green}+1 javadoc{color}.  The javadoc tool did not generate any 
warning messages.

{color:green}+1 eclipse:eclipse{color}.  The patch built with 
eclipse:eclipse.

{color:green}+1 findbugs{color}.  The patch does not introduce any new 
Findbugs (version 1.3.9) warnings.

{color:green}+1 release audit{color}.  The applied patch does not increase 
the total number of release audit warnings.

{color:green}+1 core tests{color}.  The patch passed unit tests in 
hadoop-assemblies hadoop-tools/hadoop-sls hadoop-tools/hadoop-tools-dist.

{color:green}+1 contrib tests{color}.  The patch passed contrib unit tests.

Test results: 
https://builds.apache.org/job/PreCommit-YARN-Build/1999//testReport/
Console output: https://builds.apache.org/job/PreCommit-YARN-Build/1999//console

This message is automatically generated.

 Yarn Scheduler Load Simulator
 -

 Key: YARN-1021
 URL: https://issues.apache.org/jira/browse/YARN-1021
 Project: Hadoop YARN
  Issue Type: New Feature
  Components: scheduler
Reporter: Wei Yan
Assignee: Wei Yan
 Attachments: YARN-1021-demo.tar.gz, YARN-1021-images.tar.gz, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.pdf


 The Yarn Scheduler is a fertile area of interest with different 
 implementations, e.g., Fifo, Capacity and Fair  schedulers. Meanwhile, 
 several optimizations are also made to improve scheduler performance for 
 different scenarios and workload. Each scheduler algorithm has its own set of 
 features, and drives scheduling decisions by many factors, such as fairness, 
 capacity guarantee, resource availability, etc. It is very important to 
 evaluate a scheduler algorithm very well before we deploy it in a production 
 cluster. Unfortunately, currently it is non-trivial to evaluate a scheduling 
 algorithm. Evaluating in a real cluster is always time and cost consuming, 
 and it is also very hard to find a large-enough cluster. Hence, a simulator 
 which can predict how well a scheduler algorithm for some specific workload 
 would be quite useful.
 We want to build a Scheduler Load Simulator to simulate large-scale Yarn 
 clusters and application loads in a single machine. This would be invaluable 
 in furthering Yarn by providing a tool for researchers and developers to 
 prototype new scheduler features and predict their behavior and performance 
 with reasonable amount of confidence, there-by aiding rapid innovation.
 The simulator will exercise the real Yarn ResourceManager removing the 
 network factor by simulating NodeManagers and ApplicationMasters via handling 
 and dispatching NM/AMs heartbeat events from within the same JVM.
 To keep tracking of scheduler behavior and performance, a scheduler wrapper 
 will wrap the real scheduler.
 The simulator will produce real time metrics while executing, including:
 * Resource usages for whole cluster and each queue, which can be utilized to 
 configure cluster and queue's capacity.
 * The detailed application execution trace (recorded in relation to simulated 
 time), which can be analyzed to understand/validate the  scheduler behavior 
 (individual jobs turn around time, throughput, fairness, capacity guarantee, 
 etc).
 * Several key metrics of scheduler algorithm, such as time cost of each 
 scheduler operation (allocate, handle, etc), which can be utilized by Hadoop 
 developers to find the code spots and scalability limits.
 The simulator will provide real time charts showing the behavior of the 
 scheduler and its performance.
 A short demo is available http://www.youtube.com/watch?v=6thLi8q0qLE, showing 
 how to use simulator to simulate Fair Scheduler and Capacity Scheduler.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira


[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2013-09-24 Thread Alejandro Abdelnur (JIRA)

[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13776628#comment-13776628
 ] 

Alejandro Abdelnur commented on YARN-1021:
--

+1

 Yarn Scheduler Load Simulator
 -

 Key: YARN-1021
 URL: https://issues.apache.org/jira/browse/YARN-1021
 Project: Hadoop YARN
  Issue Type: New Feature
  Components: scheduler
Reporter: Wei Yan
Assignee: Wei Yan
 Attachments: YARN-1021-demo.tar.gz, YARN-1021-images.tar.gz, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.pdf


 The Yarn Scheduler is a fertile area of interest with different 
 implementations, e.g., Fifo, Capacity and Fair  schedulers. Meanwhile, 
 several optimizations are also made to improve scheduler performance for 
 different scenarios and workload. Each scheduler algorithm has its own set of 
 features, and drives scheduling decisions by many factors, such as fairness, 
 capacity guarantee, resource availability, etc. It is very important to 
 evaluate a scheduler algorithm very well before we deploy it in a production 
 cluster. Unfortunately, currently it is non-trivial to evaluate a scheduling 
 algorithm. Evaluating in a real cluster is always time and cost consuming, 
 and it is also very hard to find a large-enough cluster. Hence, a simulator 
 which can predict how well a scheduler algorithm for some specific workload 
 would be quite useful.
 We want to build a Scheduler Load Simulator to simulate large-scale Yarn 
 clusters and application loads in a single machine. This would be invaluable 
 in furthering Yarn by providing a tool for researchers and developers to 
 prototype new scheduler features and predict their behavior and performance 
 with reasonable amount of confidence, there-by aiding rapid innovation.
 The simulator will exercise the real Yarn ResourceManager removing the 
 network factor by simulating NodeManagers and ApplicationMasters via handling 
 and dispatching NM/AMs heartbeat events from within the same JVM.
 To keep tracking of scheduler behavior and performance, a scheduler wrapper 
 will wrap the real scheduler.
 The simulator will produce real time metrics while executing, including:
 * Resource usages for whole cluster and each queue, which can be utilized to 
 configure cluster and queue's capacity.
 * The detailed application execution trace (recorded in relation to simulated 
 time), which can be analyzed to understand/validate the  scheduler behavior 
 (individual jobs turn around time, throughput, fairness, capacity guarantee, 
 etc).
 * Several key metrics of scheduler algorithm, such as time cost of each 
 scheduler operation (allocate, handle, etc), which can be utilized by Hadoop 
 developers to find the code spots and scalability limits.
 The simulator will provide real time charts showing the behavior of the 
 scheduler and its performance.
 A short demo is available http://www.youtube.com/watch?v=6thLi8q0qLE, showing 
 how to use simulator to simulate Fair Scheduler and Capacity Scheduler.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira


[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2013-09-19 Thread Alejandro Abdelnur (JIRA)

[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13771988#comment-13771988
 ] 

Alejandro Abdelnur commented on YARN-1021:
--

Thanks [~ywskycn], last set of NITs I believe:

* I still see the issue that the simulator is taking the fs.defaultFS from the 
hadoop conf instead the one set to file:/// 
* The NM simulator initialization is much faster now, still it takes it time to 
get all NM simulators running. We should print some kind of feedback to the 
terminal while this is happening, something like every 5 seconds N of M NMs 
initialized
* The documentation should mention the default port of the simulator console, 
10001
* The slsrun.sh script should not reset HADOOP_OPTS (this is handy to increase 
heap or configure to connect a remote debugger/profiler)


 Yarn Scheduler Load Simulator
 -

 Key: YARN-1021
 URL: https://issues.apache.org/jira/browse/YARN-1021
 Project: Hadoop YARN
  Issue Type: New Feature
  Components: scheduler
Reporter: Wei Yan
Assignee: Wei Yan
 Attachments: YARN-1021-demo.tar.gz, YARN-1021-images.tar.gz, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.pdf


 The Yarn Scheduler is a fertile area of interest with different 
 implementations, e.g., Fifo, Capacity and Fair  schedulers. Meanwhile, 
 several optimizations are also made to improve scheduler performance for 
 different scenarios and workload. Each scheduler algorithm has its own set of 
 features, and drives scheduling decisions by many factors, such as fairness, 
 capacity guarantee, resource availability, etc. It is very important to 
 evaluate a scheduler algorithm very well before we deploy it in a production 
 cluster. Unfortunately, currently it is non-trivial to evaluate a scheduling 
 algorithm. Evaluating in a real cluster is always time and cost consuming, 
 and it is also very hard to find a large-enough cluster. Hence, a simulator 
 which can predict how well a scheduler algorithm for some specific workload 
 would be quite useful.
 We want to build a Scheduler Load Simulator to simulate large-scale Yarn 
 clusters and application loads in a single machine. This would be invaluable 
 in furthering Yarn by providing a tool for researchers and developers to 
 prototype new scheduler features and predict their behavior and performance 
 with reasonable amount of confidence, there-by aiding rapid innovation.
 The simulator will exercise the real Yarn ResourceManager removing the 
 network factor by simulating NodeManagers and ApplicationMasters via handling 
 and dispatching NM/AMs heartbeat events from within the same JVM.
 To keep tracking of scheduler behavior and performance, a scheduler wrapper 
 will wrap the real scheduler.
 The simulator will produce real time metrics while executing, including:
 * Resource usages for whole cluster and each queue, which can be utilized to 
 configure cluster and queue's capacity.
 * The detailed application execution trace (recorded in relation to simulated 
 time), which can be analyzed to understand/validate the  scheduler behavior 
 (individual jobs turn around time, throughput, fairness, capacity guarantee, 
 etc).
 * Several key metrics of scheduler algorithm, such as time cost of each 
 scheduler operation (allocate, handle, etc), which can be utilized by Hadoop 
 developers to find the code spots and scalability limits.
 The simulator will provide real time charts showing the behavior of the 
 scheduler and its performance.
 A short demo is available http://www.youtube.com/watch?v=6thLi8q0qLE, showing 
 how to use simulator to simulate Fair Scheduler and Capacity Scheduler.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira


[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2013-09-18 Thread Hadoop QA (JIRA)

[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13771185#comment-13771185
 ] 

Hadoop QA commented on YARN-1021:
-

{color:red}-1 overall{color}.  Here are the results of testing the latest 
attachment 
  http://issues.apache.org/jira/secure/attachment/12603903/YARN-1021.patch
  against trunk revision .

{color:green}+1 @author{color}.  The patch does not contain any @author 
tags.

{color:green}+1 tests included{color}.  The patch appears to include 8 new 
or modified test files.

{color:green}+1 javac{color}.  The applied patch does not increase the 
total number of javac compiler warnings.

{color:green}+1 javadoc{color}.  The javadoc tool did not generate any 
warning messages.

{color:green}+1 eclipse:eclipse{color}.  The patch built with 
eclipse:eclipse.

{color:green}+1 findbugs{color}.  The patch does not introduce any new 
Findbugs (version 1.3.9) warnings.

{color:green}+1 release audit{color}.  The applied patch does not increase 
the total number of release audit warnings.

{color:red}-1 core tests{color}.  The patch failed these unit tests in 
hadoop-assemblies hadoop-tools/hadoop-sls hadoop-tools/hadoop-tools-dist:

  org.apache.hadoop.yarn.sls.TestSLSRunner

{color:green}+1 contrib tests{color}.  The patch passed contrib unit tests.

Test results: 
https://builds.apache.org/job/PreCommit-YARN-Build/1961//testReport/
Console output: https://builds.apache.org/job/PreCommit-YARN-Build/1961//console

This message is automatically generated.

 Yarn Scheduler Load Simulator
 -

 Key: YARN-1021
 URL: https://issues.apache.org/jira/browse/YARN-1021
 Project: Hadoop YARN
  Issue Type: New Feature
  Components: scheduler
Reporter: Wei Yan
Assignee: Wei Yan
 Attachments: YARN-1021-demo.tar.gz, YARN-1021-images.tar.gz, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.pdf


 The Yarn Scheduler is a fertile area of interest with different 
 implementations, e.g., Fifo, Capacity and Fair  schedulers. Meanwhile, 
 several optimizations are also made to improve scheduler performance for 
 different scenarios and workload. Each scheduler algorithm has its own set of 
 features, and drives scheduling decisions by many factors, such as fairness, 
 capacity guarantee, resource availability, etc. It is very important to 
 evaluate a scheduler algorithm very well before we deploy it in a production 
 cluster. Unfortunately, currently it is non-trivial to evaluate a scheduling 
 algorithm. Evaluating in a real cluster is always time and cost consuming, 
 and it is also very hard to find a large-enough cluster. Hence, a simulator 
 which can predict how well a scheduler algorithm for some specific workload 
 would be quite useful.
 We want to build a Scheduler Load Simulator to simulate large-scale Yarn 
 clusters and application loads in a single machine. This would be invaluable 
 in furthering Yarn by providing a tool for researchers and developers to 
 prototype new scheduler features and predict their behavior and performance 
 with reasonable amount of confidence, there-by aiding rapid innovation.
 The simulator will exercise the real Yarn ResourceManager removing the 
 network factor by simulating NodeManagers and ApplicationMasters via handling 
 and dispatching NM/AMs heartbeat events from within the same JVM.
 To keep tracking of scheduler behavior and performance, a scheduler wrapper 
 will wrap the real scheduler.
 The simulator will produce real time metrics while executing, including:
 * Resource usages for whole cluster and each queue, which can be utilized to 
 configure cluster and queue's capacity.
 * The detailed application execution trace (recorded in relation to simulated 
 time), which can be analyzed to understand/validate the  scheduler behavior 
 (individual jobs turn around time, throughput, fairness, capacity guarantee, 
 etc).
 * Several key metrics of scheduler algorithm, such as time cost of each 
 scheduler operation (allocate, handle, etc), which can be utilized by Hadoop 
 developers to find the code spots and scalability limits.
 The simulator will provide real time charts showing the behavior of the 
 scheduler and its performance.
 A short demo is available http://www.youtube.com/watch?v=6thLi8q0qLE, showing 
 how to use simulator to simulate Fair Scheduler and Capacity Scheduler.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira


[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2013-09-18 Thread Hadoop QA (JIRA)

[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13771269#comment-13771269
 ] 

Hadoop QA commented on YARN-1021:
-

{color:green}+1 overall{color}.  Here are the results of testing the latest 
attachment 
  http://issues.apache.org/jira/secure/attachment/12603917/YARN-1021.patch
  against trunk revision .

{color:green}+1 @author{color}.  The patch does not contain any @author 
tags.

{color:green}+1 tests included{color}.  The patch appears to include 11 new 
or modified test files.

{color:green}+1 javac{color}.  The applied patch does not increase the 
total number of javac compiler warnings.

{color:green}+1 javadoc{color}.  The javadoc tool did not generate any 
warning messages.

{color:green}+1 eclipse:eclipse{color}.  The patch built with 
eclipse:eclipse.

{color:green}+1 findbugs{color}.  The patch does not introduce any new 
Findbugs (version 1.3.9) warnings.

{color:green}+1 release audit{color}.  The applied patch does not increase 
the total number of release audit warnings.

{color:green}+1 core tests{color}.  The patch passed unit tests in 
hadoop-assemblies hadoop-tools/hadoop-sls hadoop-tools/hadoop-tools-dist.

{color:green}+1 contrib tests{color}.  The patch passed contrib unit tests.

Test results: 
https://builds.apache.org/job/PreCommit-YARN-Build/1962//testReport/
Console output: https://builds.apache.org/job/PreCommit-YARN-Build/1962//console

This message is automatically generated.

 Yarn Scheduler Load Simulator
 -

 Key: YARN-1021
 URL: https://issues.apache.org/jira/browse/YARN-1021
 Project: Hadoop YARN
  Issue Type: New Feature
  Components: scheduler
Reporter: Wei Yan
Assignee: Wei Yan
 Attachments: YARN-1021-demo.tar.gz, YARN-1021-images.tar.gz, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.pdf


 The Yarn Scheduler is a fertile area of interest with different 
 implementations, e.g., Fifo, Capacity and Fair  schedulers. Meanwhile, 
 several optimizations are also made to improve scheduler performance for 
 different scenarios and workload. Each scheduler algorithm has its own set of 
 features, and drives scheduling decisions by many factors, such as fairness, 
 capacity guarantee, resource availability, etc. It is very important to 
 evaluate a scheduler algorithm very well before we deploy it in a production 
 cluster. Unfortunately, currently it is non-trivial to evaluate a scheduling 
 algorithm. Evaluating in a real cluster is always time and cost consuming, 
 and it is also very hard to find a large-enough cluster. Hence, a simulator 
 which can predict how well a scheduler algorithm for some specific workload 
 would be quite useful.
 We want to build a Scheduler Load Simulator to simulate large-scale Yarn 
 clusters and application loads in a single machine. This would be invaluable 
 in furthering Yarn by providing a tool for researchers and developers to 
 prototype new scheduler features and predict their behavior and performance 
 with reasonable amount of confidence, there-by aiding rapid innovation.
 The simulator will exercise the real Yarn ResourceManager removing the 
 network factor by simulating NodeManagers and ApplicationMasters via handling 
 and dispatching NM/AMs heartbeat events from within the same JVM.
 To keep tracking of scheduler behavior and performance, a scheduler wrapper 
 will wrap the real scheduler.
 The simulator will produce real time metrics while executing, including:
 * Resource usages for whole cluster and each queue, which can be utilized to 
 configure cluster and queue's capacity.
 * The detailed application execution trace (recorded in relation to simulated 
 time), which can be analyzed to understand/validate the  scheduler behavior 
 (individual jobs turn around time, throughput, fairness, capacity guarantee, 
 etc).
 * Several key metrics of scheduler algorithm, such as time cost of each 
 scheduler operation (allocate, handle, etc), which can be utilized by Hadoop 
 developers to find the code spots and scalability limits.
 The simulator will provide real time charts showing the behavior of the 
 scheduler and its performance.
 A short demo is available http://www.youtube.com/watch?v=6thLi8q0qLE, showing 
 how to use simulator to simulate Fair Scheduler and Capacity Scheduler.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira


[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2013-09-17 Thread Alejandro Abdelnur (JIRA)

[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13769765#comment-13769765
 ] 

Alejandro Abdelnur commented on YARN-1021:
--

Wei, patch looks good, though some NITs that should be taken care of:

* remove LICENSE.txt/NOTICE.txt (Hadoop has those at its root)
* all the files in src/main/resources should not be there but in a 
src/main/sample-conf dir else the end up in the JAR and may be picked up 
without the user knowing
* the src/test/data dir should be in src/main/data (as it ends up in the distro)
* the test running the simulation runs for 2 mins, do we need 2 mins or can we 
make it less? like 1min or 30secs?
* hadoop-tools-dist/pom.xml, hadoop-sls entry, should not have version. there 
should be an entry for hadoop-sls with the version in the 
hadoop-project/pom.xml dependencyManagement section.
* The slsrunner should set fs.defaultFS=file:/// in the conf used to start the 
RM, else the RM will attempt to connect to HDFS.


 Yarn Scheduler Load Simulator
 -

 Key: YARN-1021
 URL: https://issues.apache.org/jira/browse/YARN-1021
 Project: Hadoop YARN
  Issue Type: New Feature
  Components: scheduler
Reporter: Wei Yan
Assignee: Wei Yan
 Attachments: YARN-1021-demo.tar.gz, YARN-1021-images.tar.gz, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.pdf


 The Yarn Scheduler is a fertile area of interest with different 
 implementations, e.g., Fifo, Capacity and Fair  schedulers. Meanwhile, 
 several optimizations are also made to improve scheduler performance for 
 different scenarios and workload. Each scheduler algorithm has its own set of 
 features, and drives scheduling decisions by many factors, such as fairness, 
 capacity guarantee, resource availability, etc. It is very important to 
 evaluate a scheduler algorithm very well before we deploy it in a production 
 cluster. Unfortunately, currently it is non-trivial to evaluate a scheduling 
 algorithm. Evaluating in a real cluster is always time and cost consuming, 
 and it is also very hard to find a large-enough cluster. Hence, a simulator 
 which can predict how well a scheduler algorithm for some specific workload 
 would be quite useful.
 We want to build a Scheduler Load Simulator to simulate large-scale Yarn 
 clusters and application loads in a single machine. This would be invaluable 
 in furthering Yarn by providing a tool for researchers and developers to 
 prototype new scheduler features and predict their behavior and performance 
 with reasonable amount of confidence, there-by aiding rapid innovation.
 The simulator will exercise the real Yarn ResourceManager removing the 
 network factor by simulating NodeManagers and ApplicationMasters via handling 
 and dispatching NM/AMs heartbeat events from within the same JVM.
 To keep tracking of scheduler behavior and performance, a scheduler wrapper 
 will wrap the real scheduler.
 The simulator will produce real time metrics while executing, including:
 * Resource usages for whole cluster and each queue, which can be utilized to 
 configure cluster and queue's capacity.
 * The detailed application execution trace (recorded in relation to simulated 
 time), which can be analyzed to understand/validate the  scheduler behavior 
 (individual jobs turn around time, throughput, fairness, capacity guarantee, 
 etc).
 * Several key metrics of scheduler algorithm, such as time cost of each 
 scheduler operation (allocate, handle, etc), which can be utilized by Hadoop 
 developers to find the code spots and scalability limits.
 The simulator will provide real time charts showing the behavior of the 
 scheduler and its performance.
 A short demo is available http://www.youtube.com/watch?v=6thLi8q0qLE, showing 
 how to use simulator to simulate Fair Scheduler and Capacity Scheduler.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira


[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2013-09-12 Thread Wei Yan (JIRA)

[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13765772#comment-13765772
 ] 

Wei Yan commented on YARN-1021:
---

Update a new patch according to [~curino]'s comments.
Main updates:
(1) Let simulator adopt rumen traces directly, using the Rumen Reader 
(JobStory). Document also updated.
(2) Add a testcase that starts the simulator using an example trace. Run 2 min 
and stop.
(3) Fix several minors (LICENSE/NOTICE, javadoc, string appends, etc).

For simulating slowstart, headroom and some other behaviors, opened jira 
YARN-1186.
For discreate event-based simulation, opened jira YARN-1187. 

 Yarn Scheduler Load Simulator
 -

 Key: YARN-1021
 URL: https://issues.apache.org/jira/browse/YARN-1021
 Project: Hadoop YARN
  Issue Type: New Feature
  Components: scheduler
Reporter: Wei Yan
Assignee: Wei Yan
 Attachments: YARN-1021-demo.tar.gz, YARN-1021-images.tar.gz, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.pdf


 The Yarn Scheduler is a fertile area of interest with different 
 implementations, e.g., Fifo, Capacity and Fair  schedulers. Meanwhile, 
 several optimizations are also made to improve scheduler performance for 
 different scenarios and workload. Each scheduler algorithm has its own set of 
 features, and drives scheduling decisions by many factors, such as fairness, 
 capacity guarantee, resource availability, etc. It is very important to 
 evaluate a scheduler algorithm very well before we deploy it in a production 
 cluster. Unfortunately, currently it is non-trivial to evaluate a scheduling 
 algorithm. Evaluating in a real cluster is always time and cost consuming, 
 and it is also very hard to find a large-enough cluster. Hence, a simulator 
 which can predict how well a scheduler algorithm for some specific workload 
 would be quite useful.
 We want to build a Scheduler Load Simulator to simulate large-scale Yarn 
 clusters and application loads in a single machine. This would be invaluable 
 in furthering Yarn by providing a tool for researchers and developers to 
 prototype new scheduler features and predict their behavior and performance 
 with reasonable amount of confidence, there-by aiding rapid innovation.
 The simulator will exercise the real Yarn ResourceManager removing the 
 network factor by simulating NodeManagers and ApplicationMasters via handling 
 and dispatching NM/AMs heartbeat events from within the same JVM.
 To keep tracking of scheduler behavior and performance, a scheduler wrapper 
 will wrap the real scheduler.
 The simulator will produce real time metrics while executing, including:
 * Resource usages for whole cluster and each queue, which can be utilized to 
 configure cluster and queue's capacity.
 * The detailed application execution trace (recorded in relation to simulated 
 time), which can be analyzed to understand/validate the  scheduler behavior 
 (individual jobs turn around time, throughput, fairness, capacity guarantee, 
 etc).
 * Several key metrics of scheduler algorithm, such as time cost of each 
 scheduler operation (allocate, handle, etc), which can be utilized by Hadoop 
 developers to find the code spots and scalability limits.
 The simulator will provide real time charts showing the behavior of the 
 scheduler and its performance.
 A short demo is available http://www.youtube.com/watch?v=6thLi8q0qLE, showing 
 how to use simulator to simulate Fair Scheduler and Capacity Scheduler.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira


[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2013-09-12 Thread Hadoop QA (JIRA)

[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13765758#comment-13765758
 ] 

Hadoop QA commented on YARN-1021:
-

{color:green}+1 overall{color}.  Here are the results of testing the latest 
attachment 
  http://issues.apache.org/jira/secure/attachment/12602846/YARN-1021.patch
  against trunk revision .

{color:green}+1 @author{color}.  The patch does not contain any @author 
tags.

{color:green}+1 tests included{color}.  The patch appears to include 6 new 
or modified test files.

{color:green}+1 javac{color}.  The applied patch does not increase the 
total number of javac compiler warnings.

{color:green}+1 javadoc{color}.  The javadoc tool did not generate any 
warning messages.

{color:green}+1 eclipse:eclipse{color}.  The patch built with 
eclipse:eclipse.

{color:green}+1 findbugs{color}.  The patch does not introduce any new 
Findbugs (version 1.3.9) warnings.

{color:green}+1 release audit{color}.  The applied patch does not increase 
the total number of release audit warnings.

{color:green}+1 core tests{color}.  The patch passed unit tests in 
hadoop-assemblies hadoop-tools/hadoop-sls hadoop-tools/hadoop-tools-dist.

{color:green}+1 contrib tests{color}.  The patch passed contrib unit tests.

Test results: 
https://builds.apache.org/job/PreCommit-YARN-Build/1903//testReport/
Console output: https://builds.apache.org/job/PreCommit-YARN-Build/1903//console

This message is automatically generated.

 Yarn Scheduler Load Simulator
 -

 Key: YARN-1021
 URL: https://issues.apache.org/jira/browse/YARN-1021
 Project: Hadoop YARN
  Issue Type: New Feature
  Components: scheduler
Reporter: Wei Yan
Assignee: Wei Yan
 Attachments: YARN-1021-demo.tar.gz, YARN-1021-images.tar.gz, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.pdf


 The Yarn Scheduler is a fertile area of interest with different 
 implementations, e.g., Fifo, Capacity and Fair  schedulers. Meanwhile, 
 several optimizations are also made to improve scheduler performance for 
 different scenarios and workload. Each scheduler algorithm has its own set of 
 features, and drives scheduling decisions by many factors, such as fairness, 
 capacity guarantee, resource availability, etc. It is very important to 
 evaluate a scheduler algorithm very well before we deploy it in a production 
 cluster. Unfortunately, currently it is non-trivial to evaluate a scheduling 
 algorithm. Evaluating in a real cluster is always time and cost consuming, 
 and it is also very hard to find a large-enough cluster. Hence, a simulator 
 which can predict how well a scheduler algorithm for some specific workload 
 would be quite useful.
 We want to build a Scheduler Load Simulator to simulate large-scale Yarn 
 clusters and application loads in a single machine. This would be invaluable 
 in furthering Yarn by providing a tool for researchers and developers to 
 prototype new scheduler features and predict their behavior and performance 
 with reasonable amount of confidence, there-by aiding rapid innovation.
 The simulator will exercise the real Yarn ResourceManager removing the 
 network factor by simulating NodeManagers and ApplicationMasters via handling 
 and dispatching NM/AMs heartbeat events from within the same JVM.
 To keep tracking of scheduler behavior and performance, a scheduler wrapper 
 will wrap the real scheduler.
 The simulator will produce real time metrics while executing, including:
 * Resource usages for whole cluster and each queue, which can be utilized to 
 configure cluster and queue's capacity.
 * The detailed application execution trace (recorded in relation to simulated 
 time), which can be analyzed to understand/validate the  scheduler behavior 
 (individual jobs turn around time, throughput, fairness, capacity guarantee, 
 etc).
 * Several key metrics of scheduler algorithm, such as time cost of each 
 scheduler operation (allocate, handle, etc), which can be utilized by Hadoop 
 developers to find the code spots and scalability limits.
 The simulator will provide real time charts showing the behavior of the 
 scheduler and its performance.
 A short demo is available http://www.youtube.com/watch?v=6thLi8q0qLE, showing 
 how to use simulator to simulate Fair Scheduler and Capacity Scheduler.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira


[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2013-08-20 Thread Carlo Curino (JIRA)

[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13745482#comment-13745482
 ] 

Carlo Curino commented on YARN-1021:


I might have missed this, but can you check whether the 
ProportionalCapacityPreemptionPolicy (running as a monitor) gets invoked 
correctly when running the simulator with the CapacityScheduler (and with 
preemption turned on)?

For the use of Clocks in the RM I think it was pretty consistent (if I remember 
correctly from our simulator attempt). Also notice that more than a faster 
version of time, it is important to achieve discrete event simulation, as this 
allows awesome debugging... just accelerating/slowing down time does not give 
you that. Please consider this seriously, as I think it would heavily increase 
the value of your simulator.

Related to that is  consistent replay (as in fully deterministic). This would 
be very very valuable to have. I can see this to be invaluable for debugging, 
testing, demonstrating features and corner cases etc. 

Simulating NM/AM is indeed costly, thought it stresses a bunch of other part of 
the system... pros and cons... I would suggest you to simply make sure that 
your architecture/design does not prevent us later on to broaden the scope of 
the simulation. 



 Yarn Scheduler Load Simulator
 -

 Key: YARN-1021
 URL: https://issues.apache.org/jira/browse/YARN-1021
 Project: Hadoop YARN
  Issue Type: New Feature
  Components: scheduler
Reporter: Wei Yan
Assignee: Wei Yan
 Attachments: YARN-1021-demo.tar.gz, YARN-1021-images.tar.gz, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.pdf


 The Yarn Scheduler is a fertile area of interest with different 
 implementations, e.g., Fifo, Capacity and Fair  schedulers. Meanwhile, 
 several optimizations are also made to improve scheduler performance for 
 different scenarios and workload. Each scheduler algorithm has its own set of 
 features, and drives scheduling decisions by many factors, such as fairness, 
 capacity guarantee, resource availability, etc. It is very important to 
 evaluate a scheduler algorithm very well before we deploy it in a production 
 cluster. Unfortunately, currently it is non-trivial to evaluate a scheduling 
 algorithm. Evaluating in a real cluster is always time and cost consuming, 
 and it is also very hard to find a large-enough cluster. Hence, a simulator 
 which can predict how well a scheduler algorithm for some specific workload 
 would be quite useful.
 We want to build a Scheduler Load Simulator to simulate large-scale Yarn 
 clusters and application loads in a single machine. This would be invaluable 
 in furthering Yarn by providing a tool for researchers and developers to 
 prototype new scheduler features and predict their behavior and performance 
 with reasonable amount of confidence, there-by aiding rapid innovation.
 The simulator will exercise the real Yarn ResourceManager removing the 
 network factor by simulating NodeManagers and ApplicationMasters via handling 
 and dispatching NM/AMs heartbeat events from within the same JVM.
 To keep tracking of scheduler behavior and performance, a scheduler wrapper 
 will wrap the real scheduler.
 The simulator will produce real time metrics while executing, including:
 * Resource usages for whole cluster and each queue, which can be utilized to 
 configure cluster and queue's capacity.
 * The detailed application execution trace (recorded in relation to simulated 
 time), which can be analyzed to understand/validate the  scheduler behavior 
 (individual jobs turn around time, throughput, fairness, capacity guarantee, 
 etc).
 * Several key metrics of scheduler algorithm, such as time cost of each 
 scheduler operation (allocate, handle, etc), which can be utilized by Hadoop 
 developers to find the code spots and scalability limits.
 The simulator will provide real time charts showing the behavior of the 
 scheduler and its performance.
 A short demo is available http://www.youtube.com/watch?v=6thLi8q0qLE, showing 
 how to use simulator to simulate Fair Scheduler and Capacity Scheduler.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira


[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2013-08-20 Thread Wei Yan (JIRA)

[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13745799#comment-13745799
 ] 

Wei Yan commented on YARN-1021:
---

[~curino] Thanks for your further comments.

bq. I might have missed this, but can you check whether the 
ProportionalCapacityPreemptionPolicy (running as a monitor) gets invoked 
correctly when running the simulator with the CapacityScheduler (and with 
preemption turned on)?

Sure, I'll check that.

bq. For the use of Clocks in the RM I think it was pretty consistent (if I 
remember correctly from our simulator attempt). Also notice that more than a 
faster version of time, it is important to achieve discrete event simulation, 
as this allows awesome debugging... just accelerating/slowing down time does 
not give you that. Please consider this seriously, as I think it would heavily 
increase the value of your simulator.

Sorry I still not understand why the simulator needs event simulation approach. 
Could you share more info about that?

bq. Related to that is consistent replay (as in fully deterministic). This 
would be very very valuable to have. I can see this to be invaluable for 
debugging, testing, demonstrating features and corner cases etc.

Yes, consistent replay is important in some cases. I'll think about it later.

bq. Simulating NM/AM is indeed costly, thought it stresses a bunch of other 
part of the system... pros and cons... I would suggest you to simply make sure 
that your architecture/design does not prevent us later on to broaden the scope 
of the simulation.

I agree that design of the simulator should conside future improvement. I'll 
update when I have some ideas.

 Yarn Scheduler Load Simulator
 -

 Key: YARN-1021
 URL: https://issues.apache.org/jira/browse/YARN-1021
 Project: Hadoop YARN
  Issue Type: New Feature
  Components: scheduler
Reporter: Wei Yan
Assignee: Wei Yan
 Attachments: YARN-1021-demo.tar.gz, YARN-1021-images.tar.gz, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.pdf


 The Yarn Scheduler is a fertile area of interest with different 
 implementations, e.g., Fifo, Capacity and Fair  schedulers. Meanwhile, 
 several optimizations are also made to improve scheduler performance for 
 different scenarios and workload. Each scheduler algorithm has its own set of 
 features, and drives scheduling decisions by many factors, such as fairness, 
 capacity guarantee, resource availability, etc. It is very important to 
 evaluate a scheduler algorithm very well before we deploy it in a production 
 cluster. Unfortunately, currently it is non-trivial to evaluate a scheduling 
 algorithm. Evaluating in a real cluster is always time and cost consuming, 
 and it is also very hard to find a large-enough cluster. Hence, a simulator 
 which can predict how well a scheduler algorithm for some specific workload 
 would be quite useful.
 We want to build a Scheduler Load Simulator to simulate large-scale Yarn 
 clusters and application loads in a single machine. This would be invaluable 
 in furthering Yarn by providing a tool for researchers and developers to 
 prototype new scheduler features and predict their behavior and performance 
 with reasonable amount of confidence, there-by aiding rapid innovation.
 The simulator will exercise the real Yarn ResourceManager removing the 
 network factor by simulating NodeManagers and ApplicationMasters via handling 
 and dispatching NM/AMs heartbeat events from within the same JVM.
 To keep tracking of scheduler behavior and performance, a scheduler wrapper 
 will wrap the real scheduler.
 The simulator will produce real time metrics while executing, including:
 * Resource usages for whole cluster and each queue, which can be utilized to 
 configure cluster and queue's capacity.
 * The detailed application execution trace (recorded in relation to simulated 
 time), which can be analyzed to understand/validate the  scheduler behavior 
 (individual jobs turn around time, throughput, fairness, capacity guarantee, 
 etc).
 * Several key metrics of scheduler algorithm, such as time cost of each 
 scheduler operation (allocate, handle, etc), which can be utilized by Hadoop 
 developers to find the code spots and scalability limits.
 The simulator will provide real time charts showing the behavior of the 
 scheduler and its performance.
 A short demo is available http://www.youtube.com/watch?v=6thLi8q0qLE, showing 
 how to use simulator to simulate Fair Scheduler and Capacity Scheduler.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira


[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2013-08-19 Thread Carlo Curino (JIRA)

[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13744493#comment-13744493
 ] 

Carlo Curino commented on YARN-1021:


Sorry for the delay. I went over the patch today together with Chris Douglas 
and here some input from the both of us.

I generally like the effort, and the live visualization is really neat. Also 
making it into a completely separate tool is convenient/safe. 

The main limitations I see in this simulator are:
* it only simulates the Scheduler code, mocking out most of the RM, and all AM 
and NM, communication, submissions... 
* If I am not mistaken runs at wall-clock time (not faster) 
* does not run the monitors which are needed for simulating preemption in the 
CapacityScheduler

An alternative approach that we explored was to hijack the Clocks around the 
RM and drive them using a discrete event simulation, thus exercising more of 
the RM code, protocols etc... and enabling faster than wall-clock speeds 
(though not trivial to achieve). We have some working but not polished code in 
this space, which we could probably provide if you think might be 
integrated/leveraged.

Ignoring alternative approaches, and broader spectrum we mentioned above, there 
are few issues with the current patch:
* It should be possible to consistently replay (seed RANDOM)
* Using Rumen reader (JobProducer, etc.) instead of parsing json directly seems 
cleaner. Also we have a synth load generator which we will release soon that 
implements the JobProducer/JobStory interface (might be nice to use that to 
drive your simulations)
* LICENSE/NOTICE should be updated to include the BSD-like licenses you bring 
in with the new libraries
* It seems somewhat hard to detect regressions w/ trunk since:
** mocks away much of the AM/NM/RM 
** few unit tests
** does not simulate important behaviors in the AM (no slow start, headroom, 
etc.)
** does not exercise failures, timeouts, etc.

Smaller issues:
* some javadoc @param unpopulated
* why a dependency on another metrics package, instead of Hadoop's?
* why NodeUpdateSchedulerEventWrapper? Doesn't seem to add anything...
* use ResourceCalculator instead of manually adjusting Resources from RR
* initMetrics is a very large method...
* SLSWebApp: is a wall of string appends. I am not very web savvy but I believe 
there should be cleaner ways to generate this. This seems hard to 
maintain/evolve.

I hope this helps. I will be traveling abroad for a couple of weeks so I might 
be slow/unresponsive. Altogether since it is rather on a side I am not too 
concern about it, the suggestions are mostly to make sure it is really useful 
and that people can use it / maintain it overtime. If committed as is will do 
no harm, but I think it risk to be dropped in, used twice for FairScheduler 
work, and than loose relevance and get out of sync from trunk.


 Yarn Scheduler Load Simulator
 -

 Key: YARN-1021
 URL: https://issues.apache.org/jira/browse/YARN-1021
 Project: Hadoop YARN
  Issue Type: New Feature
  Components: scheduler
Reporter: Wei Yan
Assignee: Wei Yan
 Attachments: YARN-1021-demo.tar.gz, YARN-1021-images.tar.gz, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.pdf


 The Yarn Scheduler is a fertile area of interest with different 
 implementations, e.g., Fifo, Capacity and Fair  schedulers. Meanwhile, 
 several optimizations are also made to improve scheduler performance for 
 different scenarios and workload. Each scheduler algorithm has its own set of 
 features, and drives scheduling decisions by many factors, such as fairness, 
 capacity guarantee, resource availability, etc. It is very important to 
 evaluate a scheduler algorithm very well before we deploy it in a production 
 cluster. Unfortunately, currently it is non-trivial to evaluate a scheduling 
 algorithm. Evaluating in a real cluster is always time and cost consuming, 
 and it is also very hard to find a large-enough cluster. Hence, a simulator 
 which can predict how well a scheduler algorithm for some specific workload 
 would be quite useful.
 We want to build a Scheduler Load Simulator to simulate large-scale Yarn 
 clusters and application loads in a single machine. This would be invaluable 
 in furthering Yarn by providing a tool for researchers and developers to 
 prototype new scheduler features and predict their behavior and performance 
 with reasonable amount of confidence, there-by aiding rapid innovation.
 The simulator will exercise the real Yarn ResourceManager removing the 
 network factor by simulating NodeManagers and ApplicationMasters via handling 
 and dispatching NM/AMs heartbeat events from within the same JVM.
 To keep tracking of scheduler 

[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2013-08-13 Thread Wei Yan (JIRA)

[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13738556#comment-13738556
 ] 

Wei Yan commented on YARN-1021:
---

Updates of the patch: Reduce the number of threads needed for NMSimulators. 
Before, each NMSimulator uses one thread (for its AsyncDispatcher). Currently 
removed AsyncDispatcher and the total number of threads needed only depends on 
the thread pool size.

 Yarn Scheduler Load Simulator
 -

 Key: YARN-1021
 URL: https://issues.apache.org/jira/browse/YARN-1021
 Project: Hadoop YARN
  Issue Type: New Feature
  Components: scheduler
Reporter: Wei Yan
Assignee: Wei Yan
 Attachments: YARN-1021-demo.tar.gz, YARN-1021-images.tar.gz, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.pdf


 The Yarn Scheduler is a fertile area of interest with different 
 implementations, e.g., Fifo, Capacity and Fair  schedulers. Meanwhile, 
 several optimizations are also made to improve scheduler performance for 
 different scenarios and workload. Each scheduler algorithm has its own set of 
 features, and drives scheduling decisions by many factors, such as fairness, 
 capacity guarantee, resource availability, etc. It is very important to 
 evaluate a scheduler algorithm very well before we deploy it in a production 
 cluster. Unfortunately, currently it is non-trivial to evaluate a scheduling 
 algorithm. Evaluating in a real cluster is always time and cost consuming, 
 and it is also very hard to find a large-enough cluster. Hence, a simulator 
 which can predict how well a scheduler algorithm for some specific workload 
 would be quite useful.
 We want to build a Scheduler Load Simulator to simulate large-scale Yarn 
 clusters and application loads in a single machine. This would be invaluable 
 in furthering Yarn by providing a tool for researchers and developers to 
 prototype new scheduler features and predict their behavior and performance 
 with reasonable amount of confidence, there-by aiding rapid innovation.
 The simulator will exercise the real Yarn ResourceManager removing the 
 network factor by simulating NodeManagers and ApplicationMasters via handling 
 and dispatching NM/AMs heartbeat events from within the same JVM.
 To keep tracking of scheduler behavior and performance, a scheduler wrapper 
 will wrap the real scheduler.
 The simulator will produce real time metrics while executing, including:
 * Resource usages for whole cluster and each queue, which can be utilized to 
 configure cluster and queue's capacity.
 * The detailed application execution trace (recorded in relation to simulated 
 time), which can be analyzed to understand/validate the  scheduler behavior 
 (individual jobs turn around time, throughput, fairness, capacity guarantee, 
 etc).
 * Several key metrics of scheduler algorithm, such as time cost of each 
 scheduler operation (allocate, handle, etc), which can be utilized by Hadoop 
 developers to find the code spots and scalability limits.
 The simulator will provide real time charts showing the behavior of the 
 scheduler and its performance.
 A short demo is available http://www.youtube.com/watch?v=6thLi8q0qLE, showing 
 how to use simulator to simulate Fair Scheduler and Capacity Scheduler.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira


[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2013-08-13 Thread Hadoop QA (JIRA)

[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13738570#comment-13738570
 ] 

Hadoop QA commented on YARN-1021:
-

{color:green}+1 overall{color}.  Here are the results of testing the latest 
attachment 
  http://issues.apache.org/jira/secure/attachment/12597774/YARN-1021.patch
  against trunk revision .

{color:green}+1 @author{color}.  The patch does not contain any @author 
tags.

{color:green}+1 tests included{color}.  The patch appears to include 4 new 
or modified test files.

{color:green}+1 javac{color}.  The applied patch does not increase the 
total number of javac compiler warnings.

{color:green}+1 javadoc{color}.  The javadoc tool did not generate any 
warning messages.

{color:green}+1 eclipse:eclipse{color}.  The patch built with 
eclipse:eclipse.

{color:green}+1 findbugs{color}.  The patch does not introduce any new 
Findbugs (version 1.3.9) warnings.

{color:green}+1 release audit{color}.  The applied patch does not increase 
the total number of release audit warnings.

{color:green}+1 core tests{color}.  The patch passed unit tests in 
hadoop-assemblies hadoop-tools/hadoop-sls hadoop-tools/hadoop-tools-dist.

{color:green}+1 contrib tests{color}.  The patch passed contrib unit tests.

Test results: 
https://builds.apache.org/job/PreCommit-YARN-Build/1705//testReport/
Console output: https://builds.apache.org/job/PreCommit-YARN-Build/1705//console

This message is automatically generated.

 Yarn Scheduler Load Simulator
 -

 Key: YARN-1021
 URL: https://issues.apache.org/jira/browse/YARN-1021
 Project: Hadoop YARN
  Issue Type: New Feature
  Components: scheduler
Reporter: Wei Yan
Assignee: Wei Yan
 Attachments: YARN-1021-demo.tar.gz, YARN-1021-images.tar.gz, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.pdf


 The Yarn Scheduler is a fertile area of interest with different 
 implementations, e.g., Fifo, Capacity and Fair  schedulers. Meanwhile, 
 several optimizations are also made to improve scheduler performance for 
 different scenarios and workload. Each scheduler algorithm has its own set of 
 features, and drives scheduling decisions by many factors, such as fairness, 
 capacity guarantee, resource availability, etc. It is very important to 
 evaluate a scheduler algorithm very well before we deploy it in a production 
 cluster. Unfortunately, currently it is non-trivial to evaluate a scheduling 
 algorithm. Evaluating in a real cluster is always time and cost consuming, 
 and it is also very hard to find a large-enough cluster. Hence, a simulator 
 which can predict how well a scheduler algorithm for some specific workload 
 would be quite useful.
 We want to build a Scheduler Load Simulator to simulate large-scale Yarn 
 clusters and application loads in a single machine. This would be invaluable 
 in furthering Yarn by providing a tool for researchers and developers to 
 prototype new scheduler features and predict their behavior and performance 
 with reasonable amount of confidence, there-by aiding rapid innovation.
 The simulator will exercise the real Yarn ResourceManager removing the 
 network factor by simulating NodeManagers and ApplicationMasters via handling 
 and dispatching NM/AMs heartbeat events from within the same JVM.
 To keep tracking of scheduler behavior and performance, a scheduler wrapper 
 will wrap the real scheduler.
 The simulator will produce real time metrics while executing, including:
 * Resource usages for whole cluster and each queue, which can be utilized to 
 configure cluster and queue's capacity.
 * The detailed application execution trace (recorded in relation to simulated 
 time), which can be analyzed to understand/validate the  scheduler behavior 
 (individual jobs turn around time, throughput, fairness, capacity guarantee, 
 etc).
 * Several key metrics of scheduler algorithm, such as time cost of each 
 scheduler operation (allocate, handle, etc), which can be utilized by Hadoop 
 developers to find the code spots and scalability limits.
 The simulator will provide real time charts showing the behavior of the 
 scheduler and its performance.
 A short demo is available http://www.youtube.com/watch?v=6thLi8q0qLE, showing 
 how to use simulator to simulate Fair Scheduler and Capacity Scheduler.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira


[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2013-08-09 Thread Alejandro Abdelnur (JIRA)

[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13735170#comment-13735170
 ] 

Alejandro Abdelnur commented on YARN-1021:
--

Patch looks good, some comments:

Documentation:

* 'Usage': you are assuming the user built hadoop, instead, you should explain 
things assuming the user is standing in the hadoop install directory, after 
they expand the tarball, standing where share/ and etc/.

* 'Usage step 1': mention the sls-runner.xml has all the default values for 
this properties (it should if not). Also, mention that the hadoop RM/scheduler 
configuration will be taken from the Hadoop conf directory.

* 'Usage step 2': use path to rumen2sls.sh from install directory.

* 'Usage step 3': use path to slsrun.sh from install directory.

Code:

* AMSimulator: var definitions, one var def per line, don't comma them.
* AMSimulator: middleStep() method, no need to use 'this.' when invoking 
methods.
* AMSimulator: instead using SLF4J MessageFormatter, use JDK MessageFormat as 
in other places (note you'll have to change from {} to {#} in the templates).
* ResourceSchedulerWrapper: instead using getRuntime() to add a shutdown hook 
it should use Hadoop's ShutdownHookManager

html/simulateTraces.html should be renamed to showSimulationTrace.html. The 
documentation should explain clearly that a SLS run produces a simulation trace 
that can be seen after running the simulation.


 Yarn Scheduler Load Simulator
 -

 Key: YARN-1021
 URL: https://issues.apache.org/jira/browse/YARN-1021
 Project: Hadoop YARN
  Issue Type: New Feature
  Components: scheduler
Reporter: Wei Yan
Assignee: Wei Yan
 Attachments: YARN-1021-demo.tar.gz, YARN-1021-images.tar.gz, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.pdf


 The Yarn Scheduler is a fertile area of interest with different 
 implementations, e.g., Fifo, Capacity and Fair  schedulers. Meanwhile, 
 several optimizations are also made to improve scheduler performance for 
 different scenarios and workload. Each scheduler algorithm has its own set of 
 features, and drives scheduling decisions by many factors, such as fairness, 
 capacity guarantee, resource availability, etc. It is very important to 
 evaluate a scheduler algorithm very well before we deploy it in a production 
 cluster. Unfortunately, currently it is non-trivial to evaluate a scheduling 
 algorithm. Evaluating in a real cluster is always time and cost consuming, 
 and it is also very hard to find a large-enough cluster. Hence, a simulator 
 which can predict how well a scheduler algorithm for some specific workload 
 would be quite useful.
 We want to build a Scheduler Load Simulator to simulate large-scale Yarn 
 clusters and application loads in a single machine. This would be invaluable 
 in furthering Yarn by providing a tool for researchers and developers to 
 prototype new scheduler features and predict their behavior and performance 
 with reasonable amount of confidence, there-by aiding rapid innovation.
 The simulator will exercise the real Yarn ResourceManager removing the 
 network factor by simulating NodeManagers and ApplicationMasters via handling 
 and dispatching NM/AMs heartbeat events from within the same JVM.
 To keep tracking of scheduler behavior and performance, a scheduler wrapper 
 will wrap the real scheduler.
 The simulator will produce real time metrics while executing, including:
 * Resource usages for whole cluster and each queue, which can be utilized to 
 configure cluster and queue's capacity.
 * The detailed application execution trace (recorded in relation to simulated 
 time), which can be analyzed to understand/validate the  scheduler behavior 
 (individual jobs turn around time, throughput, fairness, capacity guarantee, 
 etc).
 * Several key metrics of scheduler algorithm, such as time cost of each 
 scheduler operation (allocate, handle, etc), which can be utilized by Hadoop 
 developers to find the code spots and scalability limits.
 The simulator will provide real time charts showing the behavior of the 
 scheduler and its performance.
 A short demo is available http://www.youtube.com/watch?v=6thLi8q0qLE, showing 
 how to use simulator to simulate Fair Scheduler and Capacity Scheduler.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira


[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2013-08-09 Thread Hadoop QA (JIRA)

[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13735649#comment-13735649
 ] 

Hadoop QA commented on YARN-1021:
-

{color:green}+1 overall{color}.  Here are the results of testing the latest 
attachment 
  http://issues.apache.org/jira/secure/attachment/12597233/YARN-1021.patch
  against trunk revision .

{color:green}+1 @author{color}.  The patch does not contain any @author 
tags.

{color:green}+1 tests included{color}.  The patch appears to include 4 new 
or modified test files.

{color:green}+1 javac{color}.  The applied patch does not increase the 
total number of javac compiler warnings.

{color:green}+1 javadoc{color}.  The javadoc tool did not generate any 
warning messages.

{color:green}+1 eclipse:eclipse{color}.  The patch built with 
eclipse:eclipse.

{color:green}+1 findbugs{color}.  The patch does not introduce any new 
Findbugs (version 1.3.9) warnings.

{color:green}+1 release audit{color}.  The applied patch does not increase 
the total number of release audit warnings.

{color:green}+1 core tests{color}.  The patch passed unit tests in 
hadoop-assemblies hadoop-tools/hadoop-sls hadoop-tools/hadoop-tools-dist.

{color:green}+1 contrib tests{color}.  The patch passed contrib unit tests.

Test results: 
https://builds.apache.org/job/PreCommit-YARN-Build/1688//testReport/
Console output: https://builds.apache.org/job/PreCommit-YARN-Build/1688//console

This message is automatically generated.

 Yarn Scheduler Load Simulator
 -

 Key: YARN-1021
 URL: https://issues.apache.org/jira/browse/YARN-1021
 Project: Hadoop YARN
  Issue Type: New Feature
  Components: scheduler
Reporter: Wei Yan
Assignee: Wei Yan
 Attachments: YARN-1021-demo.tar.gz, YARN-1021-images.tar.gz, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.pdf


 The Yarn Scheduler is a fertile area of interest with different 
 implementations, e.g., Fifo, Capacity and Fair  schedulers. Meanwhile, 
 several optimizations are also made to improve scheduler performance for 
 different scenarios and workload. Each scheduler algorithm has its own set of 
 features, and drives scheduling decisions by many factors, such as fairness, 
 capacity guarantee, resource availability, etc. It is very important to 
 evaluate a scheduler algorithm very well before we deploy it in a production 
 cluster. Unfortunately, currently it is non-trivial to evaluate a scheduling 
 algorithm. Evaluating in a real cluster is always time and cost consuming, 
 and it is also very hard to find a large-enough cluster. Hence, a simulator 
 which can predict how well a scheduler algorithm for some specific workload 
 would be quite useful.
 We want to build a Scheduler Load Simulator to simulate large-scale Yarn 
 clusters and application loads in a single machine. This would be invaluable 
 in furthering Yarn by providing a tool for researchers and developers to 
 prototype new scheduler features and predict their behavior and performance 
 with reasonable amount of confidence, there-by aiding rapid innovation.
 The simulator will exercise the real Yarn ResourceManager removing the 
 network factor by simulating NodeManagers and ApplicationMasters via handling 
 and dispatching NM/AMs heartbeat events from within the same JVM.
 To keep tracking of scheduler behavior and performance, a scheduler wrapper 
 will wrap the real scheduler.
 The simulator will produce real time metrics while executing, including:
 * Resource usages for whole cluster and each queue, which can be utilized to 
 configure cluster and queue's capacity.
 * The detailed application execution trace (recorded in relation to simulated 
 time), which can be analyzed to understand/validate the  scheduler behavior 
 (individual jobs turn around time, throughput, fairness, capacity guarantee, 
 etc).
 * Several key metrics of scheduler algorithm, such as time cost of each 
 scheduler operation (allocate, handle, etc), which can be utilized by Hadoop 
 developers to find the code spots and scalability limits.
 The simulator will provide real time charts showing the behavior of the 
 scheduler and its performance.
 A short demo is available http://www.youtube.com/watch?v=6thLi8q0qLE, showing 
 how to use simulator to simulate Fair Scheduler and Capacity Scheduler.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira


[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2013-08-08 Thread Alejandro Abdelnur (JIRA)

[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13734221#comment-13734221
 ] 

Alejandro Abdelnur commented on YARN-1021:
--

Wei, first of all, nice.

I'm not convinced (even if I suggested that to you offline) on having dirs 
under share/hadoop/tools/sls/conf being 'configurable' and added to the 
classpath.

Instead I would suggest the following:

The stuff under share/hadoop/tools/sls/ should be samples, i.e.: sample-conf/  
sample-data

The runmen2sls.sh  slsrunner.sh scripts should not add the sample-conf dir to 
the classpath, they should just add the JARs.

And the documentation should state that sample-conf/ files should be copied to 
the hadoop conf/ directory to run the simulator.




 Yarn Scheduler Load Simulator
 -

 Key: YARN-1021
 URL: https://issues.apache.org/jira/browse/YARN-1021
 Project: Hadoop YARN
  Issue Type: New Feature
  Components: scheduler
Reporter: Wei Yan
Assignee: Wei Yan
 Attachments: YARN-1021-demo.tar.gz, YARN-1021-images.tar.gz, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.pdf


 The Yarn Scheduler is a fertile area of interest with different 
 implementations, e.g., Fifo, Capacity and Fair  schedulers. Meanwhile, 
 several optimizations are also made to improve scheduler performance for 
 different scenarios and workload. Each scheduler algorithm has its own set of 
 features, and drives scheduling decisions by many factors, such as fairness, 
 capacity guarantee, resource availability, etc. It is very important to 
 evaluate a scheduler algorithm very well before we deploy it in a production 
 cluster. Unfortunately, currently it is non-trivial to evaluate a scheduling 
 algorithm. Evaluating in a real cluster is always time and cost consuming, 
 and it is also very hard to find a large-enough cluster. Hence, a simulator 
 which can predict how well a scheduler algorithm for some specific workload 
 would be quite useful.
 We want to build a Scheduler Load Simulator to simulate large-scale Yarn 
 clusters and application loads in a single machine. This would be invaluable 
 in furthering Yarn by providing a tool for researchers and developers to 
 prototype new scheduler features and predict their behavior and performance 
 with reasonable amount of confidence, there-by aiding rapid innovation.
 The simulator will exercise the real Yarn ResourceManager removing the 
 network factor by simulating NodeManagers and ApplicationMasters via handling 
 and dispatching NM/AMs heartbeat events from within the same JVM.
 To keep tracking of scheduler behavior and performance, a scheduler wrapper 
 will wrap the real scheduler.
 The simulator will produce real time metrics while executing, including:
 * Resource usages for whole cluster and each queue, which can be utilized to 
 configure cluster and queue's capacity.
 * The detailed application execution trace (recorded in relation to simulated 
 time), which can be analyzed to understand/validate the  scheduler behavior 
 (individual jobs turn around time, throughput, fairness, capacity guarantee, 
 etc).
 * Several key metrics of scheduler algorithm, such as time cost of each 
 scheduler operation (allocate, handle, etc), which can be utilized by Hadoop 
 developers to find the code spots and scalability limits.
 The simulator will provide real time charts showing the behavior of the 
 scheduler and its performance.
 A short demo is available http://www.youtube.com/watch?v=6thLi8q0qLE, showing 
 how to use simulator to simulate Fair Scheduler and Capacity Scheduler.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira


[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2013-08-08 Thread Hadoop QA (JIRA)

[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13734347#comment-13734347
 ] 

Hadoop QA commented on YARN-1021:
-

{color:green}+1 overall{color}.  Here are the results of testing the latest 
attachment 
  http://issues.apache.org/jira/secure/attachment/12597013/YARN-1021.patch
  against trunk revision .

{color:green}+1 @author{color}.  The patch does not contain any @author 
tags.

{color:green}+1 tests included{color}.  The patch appears to include 4 new 
or modified test files.

{color:green}+1 javac{color}.  The applied patch does not increase the 
total number of javac compiler warnings.

{color:green}+1 javadoc{color}.  The javadoc tool did not generate any 
warning messages.

{color:green}+1 eclipse:eclipse{color}.  The patch built with 
eclipse:eclipse.

{color:green}+1 findbugs{color}.  The patch does not introduce any new 
Findbugs (version 1.3.9) warnings.

{color:green}+1 release audit{color}.  The applied patch does not increase 
the total number of release audit warnings.

{color:green}+1 core tests{color}.  The patch passed unit tests in 
hadoop-assemblies hadoop-tools/hadoop-sls hadoop-tools/hadoop-tools-dist.

{color:green}+1 contrib tests{color}.  The patch passed contrib unit tests.

Test results: 
https://builds.apache.org/job/PreCommit-YARN-Build/1680//testReport/
Console output: https://builds.apache.org/job/PreCommit-YARN-Build/1680//console

This message is automatically generated.

 Yarn Scheduler Load Simulator
 -

 Key: YARN-1021
 URL: https://issues.apache.org/jira/browse/YARN-1021
 Project: Hadoop YARN
  Issue Type: New Feature
  Components: scheduler
Reporter: Wei Yan
Assignee: Wei Yan
 Attachments: YARN-1021-demo.tar.gz, YARN-1021-images.tar.gz, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.patch, YARN-1021.pdf


 The Yarn Scheduler is a fertile area of interest with different 
 implementations, e.g., Fifo, Capacity and Fair  schedulers. Meanwhile, 
 several optimizations are also made to improve scheduler performance for 
 different scenarios and workload. Each scheduler algorithm has its own set of 
 features, and drives scheduling decisions by many factors, such as fairness, 
 capacity guarantee, resource availability, etc. It is very important to 
 evaluate a scheduler algorithm very well before we deploy it in a production 
 cluster. Unfortunately, currently it is non-trivial to evaluate a scheduling 
 algorithm. Evaluating in a real cluster is always time and cost consuming, 
 and it is also very hard to find a large-enough cluster. Hence, a simulator 
 which can predict how well a scheduler algorithm for some specific workload 
 would be quite useful.
 We want to build a Scheduler Load Simulator to simulate large-scale Yarn 
 clusters and application loads in a single machine. This would be invaluable 
 in furthering Yarn by providing a tool for researchers and developers to 
 prototype new scheduler features and predict their behavior and performance 
 with reasonable amount of confidence, there-by aiding rapid innovation.
 The simulator will exercise the real Yarn ResourceManager removing the 
 network factor by simulating NodeManagers and ApplicationMasters via handling 
 and dispatching NM/AMs heartbeat events from within the same JVM.
 To keep tracking of scheduler behavior and performance, a scheduler wrapper 
 will wrap the real scheduler.
 The simulator will produce real time metrics while executing, including:
 * Resource usages for whole cluster and each queue, which can be utilized to 
 configure cluster and queue's capacity.
 * The detailed application execution trace (recorded in relation to simulated 
 time), which can be analyzed to understand/validate the  scheduler behavior 
 (individual jobs turn around time, throughput, fairness, capacity guarantee, 
 etc).
 * Several key metrics of scheduler algorithm, such as time cost of each 
 scheduler operation (allocate, handle, etc), which can be utilized by Hadoop 
 developers to find the code spots and scalability limits.
 The simulator will provide real time charts showing the behavior of the 
 scheduler and its performance.
 A short demo is available http://www.youtube.com/watch?v=6thLi8q0qLE, showing 
 how to use simulator to simulate Fair Scheduler and Capacity Scheduler.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira


[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2013-08-07 Thread Hadoop QA (JIRA)

[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13732774#comment-13732774
 ] 

Hadoop QA commented on YARN-1021:
-

{color:red}-1 overall{color}.  Here are the results of testing the latest 
attachment 
  http://issues.apache.org/jira/secure/attachment/12596714/YARN-1021.patch
  against trunk revision .

{color:green}+1 @author{color}.  The patch does not contain any @author 
tags.

{color:green}+1 tests included{color}.  The patch appears to include 4 new 
or modified test files.

{color:green}+1 javac{color}.  The applied patch does not increase the 
total number of javac compiler warnings.

{color:green}+1 javadoc{color}.  The javadoc tool did not generate any 
warning messages.

{color:green}+1 eclipse:eclipse{color}.  The patch built with 
eclipse:eclipse.

{color:red}-1 findbugs{color}.  The patch appears to cause Findbugs 
(version 1.3.9) to fail.

{color:red}-1 release audit{color}.  The applied patch generated 4 
release audit warnings.

{color:green}+1 core tests{color}.  The patch passed unit tests in 
hadoop-assemblies hadoop-tools/hadoop-sls hadoop-tools/hadoop-tools-dist.

{color:green}+1 contrib tests{color}.  The patch passed contrib unit tests.

Test results: 
https://builds.apache.org/job/PreCommit-YARN-Build/1668//testReport/
Release audit warnings: 
https://builds.apache.org/job/PreCommit-YARN-Build/1668//artifact/trunk/patchprocess/patchReleaseAuditProblems.txt
Console output: https://builds.apache.org/job/PreCommit-YARN-Build/1668//console

This message is automatically generated.

 Yarn Scheduler Load Simulator
 -

 Key: YARN-1021
 URL: https://issues.apache.org/jira/browse/YARN-1021
 Project: Hadoop YARN
  Issue Type: New Feature
  Components: scheduler
Reporter: Wei Yan
Assignee: Wei Yan
 Attachments: YARN-1021-demo.tar.gz, YARN-1021-images.tar.gz, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.pdf


 The Yarn Scheduler is a fertile area of interest with different 
 implementations, e.g., Fifo, Capacity and Fair  schedulers. Meanwhile, 
 several optimizations are also made to improve scheduler performance for 
 different scenarios and workload. Each scheduler algorithm has its own set of 
 features, and drives scheduling decisions by many factors, such as fairness, 
 capacity guarantee, resource availability, etc. It is very important to 
 evaluate a scheduler algorithm very well before we deploy it in a production 
 cluster. Unfortunately, currently it is non-trivial to evaluate a scheduling 
 algorithm. Evaluating in a real cluster is always time and cost consuming, 
 and it is also very hard to find a large-enough cluster. Hence, a simulator 
 which can predict how well a scheduler algorithm for some specific workload 
 would be quite useful.
 We want to build a Scheduler Load Simulator to simulate large-scale Yarn 
 clusters and application loads in a single machine. This would be invaluable 
 in furthering Yarn by providing a tool for researchers and developers to 
 prototype new scheduler features and predict their behavior and performance 
 with reasonable amount of confidence, there-by aiding rapid innovation.
 The simulator will exercise the real Yarn ResourceManager removing the 
 network factor by simulating NodeManagers and ApplicationMasters via handling 
 and dispatching NM/AMs heartbeat events from within the same JVM.
 To keep tracking of scheduler behavior and performance, a scheduler wrapper 
 will wrap the real scheduler.
 The simulator will produce real time metrics while executing, including:
 * Resource usages for whole cluster and each queue, which can be utilized to 
 configure cluster and queue's capacity.
 * The detailed application execution trace (recorded in relation to simulated 
 time), which can be analyzed to understand/validate the  scheduler behavior 
 (individual jobs turn around time, throughput, fairness, capacity guarantee, 
 etc).
 * Several key metrics of scheduler algorithm, such as time cost of each 
 scheduler operation (allocate, handle, etc), which can be utilized by Hadoop 
 developers to find the code spots and scalability limits.
 The simulator will provide real time charts showing the behavior of the 
 scheduler and its performance.
 A short demo is available http://www.youtube.com/watch?v=6thLi8q0qLE, showing 
 how to use simulator to simulate Fair Scheduler and Capacity Scheduler.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira


[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2013-08-07 Thread Hadoop QA (JIRA)

[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13732829#comment-13732829
 ] 

Hadoop QA commented on YARN-1021:
-

{color:red}-1 overall{color}.  Here are the results of testing the latest 
attachment 
  http://issues.apache.org/jira/secure/attachment/12596724/YARN-1021.patch
  against trunk revision .

{color:green}+1 @author{color}.  The patch does not contain any @author 
tags.

{color:green}+1 tests included{color}.  The patch appears to include 4 new 
or modified test files.

{color:green}+1 javac{color}.  The applied patch does not increase the 
total number of javac compiler warnings.

{color:green}+1 javadoc{color}.  The javadoc tool did not generate any 
warning messages.

{color:green}+1 eclipse:eclipse{color}.  The patch built with 
eclipse:eclipse.

{color:green}+1 findbugs{color}.  The patch does not introduce any new 
Findbugs (version 1.3.9) warnings.

{color:red}-1 release audit{color}.  The applied patch generated 4 
release audit warnings.

{color:green}+1 core tests{color}.  The patch passed unit tests in 
hadoop-assemblies hadoop-tools/hadoop-sls hadoop-tools/hadoop-tools-dist.

{color:green}+1 contrib tests{color}.  The patch passed contrib unit tests.

Test results: 
https://builds.apache.org/job/PreCommit-YARN-Build/1670//testReport/
Release audit warnings: 
https://builds.apache.org/job/PreCommit-YARN-Build/1670//artifact/trunk/patchprocess/patchReleaseAuditProblems.txt
Console output: https://builds.apache.org/job/PreCommit-YARN-Build/1670//console

This message is automatically generated.

 Yarn Scheduler Load Simulator
 -

 Key: YARN-1021
 URL: https://issues.apache.org/jira/browse/YARN-1021
 Project: Hadoop YARN
  Issue Type: New Feature
  Components: scheduler
Reporter: Wei Yan
Assignee: Wei Yan
 Attachments: YARN-1021-demo.tar.gz, YARN-1021-images.tar.gz, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.pdf


 The Yarn Scheduler is a fertile area of interest with different 
 implementations, e.g., Fifo, Capacity and Fair  schedulers. Meanwhile, 
 several optimizations are also made to improve scheduler performance for 
 different scenarios and workload. Each scheduler algorithm has its own set of 
 features, and drives scheduling decisions by many factors, such as fairness, 
 capacity guarantee, resource availability, etc. It is very important to 
 evaluate a scheduler algorithm very well before we deploy it in a production 
 cluster. Unfortunately, currently it is non-trivial to evaluate a scheduling 
 algorithm. Evaluating in a real cluster is always time and cost consuming, 
 and it is also very hard to find a large-enough cluster. Hence, a simulator 
 which can predict how well a scheduler algorithm for some specific workload 
 would be quite useful.
 We want to build a Scheduler Load Simulator to simulate large-scale Yarn 
 clusters and application loads in a single machine. This would be invaluable 
 in furthering Yarn by providing a tool for researchers and developers to 
 prototype new scheduler features and predict their behavior and performance 
 with reasonable amount of confidence, there-by aiding rapid innovation.
 The simulator will exercise the real Yarn ResourceManager removing the 
 network factor by simulating NodeManagers and ApplicationMasters via handling 
 and dispatching NM/AMs heartbeat events from within the same JVM.
 To keep tracking of scheduler behavior and performance, a scheduler wrapper 
 will wrap the real scheduler.
 The simulator will produce real time metrics while executing, including:
 * Resource usages for whole cluster and each queue, which can be utilized to 
 configure cluster and queue's capacity.
 * The detailed application execution trace (recorded in relation to simulated 
 time), which can be analyzed to understand/validate the  scheduler behavior 
 (individual jobs turn around time, throughput, fairness, capacity guarantee, 
 etc).
 * Several key metrics of scheduler algorithm, such as time cost of each 
 scheduler operation (allocate, handle, etc), which can be utilized by Hadoop 
 developers to find the code spots and scalability limits.
 The simulator will provide real time charts showing the behavior of the 
 scheduler and its performance.
 A short demo is available http://www.youtube.com/watch?v=6thLi8q0qLE, showing 
 how to use simulator to simulate Fair Scheduler and Capacity Scheduler.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira


[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2013-08-07 Thread Hadoop QA (JIRA)

[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13732873#comment-13732873
 ] 

Hadoop QA commented on YARN-1021:
-

{color:green}+1 overall{color}.  Here are the results of testing the latest 
attachment 
  http://issues.apache.org/jira/secure/attachment/12596733/YARN-1021.patch
  against trunk revision .

{color:green}+1 @author{color}.  The patch does not contain any @author 
tags.

{color:green}+1 tests included{color}.  The patch appears to include 4 new 
or modified test files.

{color:green}+1 javac{color}.  The applied patch does not increase the 
total number of javac compiler warnings.

{color:green}+1 javadoc{color}.  The javadoc tool did not generate any 
warning messages.

{color:green}+1 eclipse:eclipse{color}.  The patch built with 
eclipse:eclipse.

{color:green}+1 findbugs{color}.  The patch does not introduce any new 
Findbugs (version 1.3.9) warnings.

{color:green}+1 release audit{color}.  The applied patch does not increase 
the total number of release audit warnings.

{color:green}+1 core tests{color}.  The patch passed unit tests in 
hadoop-assemblies hadoop-tools/hadoop-sls hadoop-tools/hadoop-tools-dist.

{color:green}+1 contrib tests{color}.  The patch passed contrib unit tests.

Test results: 
https://builds.apache.org/job/PreCommit-YARN-Build/1672//testReport/
Console output: https://builds.apache.org/job/PreCommit-YARN-Build/1672//console

This message is automatically generated.

 Yarn Scheduler Load Simulator
 -

 Key: YARN-1021
 URL: https://issues.apache.org/jira/browse/YARN-1021
 Project: Hadoop YARN
  Issue Type: New Feature
  Components: scheduler
Reporter: Wei Yan
Assignee: Wei Yan
 Attachments: YARN-1021-demo.tar.gz, YARN-1021-images.tar.gz, 
 YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, YARN-1021.patch, 
 YARN-1021.pdf


 The Yarn Scheduler is a fertile area of interest with different 
 implementations, e.g., Fifo, Capacity and Fair  schedulers. Meanwhile, 
 several optimizations are also made to improve scheduler performance for 
 different scenarios and workload. Each scheduler algorithm has its own set of 
 features, and drives scheduling decisions by many factors, such as fairness, 
 capacity guarantee, resource availability, etc. It is very important to 
 evaluate a scheduler algorithm very well before we deploy it in a production 
 cluster. Unfortunately, currently it is non-trivial to evaluate a scheduling 
 algorithm. Evaluating in a real cluster is always time and cost consuming, 
 and it is also very hard to find a large-enough cluster. Hence, a simulator 
 which can predict how well a scheduler algorithm for some specific workload 
 would be quite useful.
 We want to build a Scheduler Load Simulator to simulate large-scale Yarn 
 clusters and application loads in a single machine. This would be invaluable 
 in furthering Yarn by providing a tool for researchers and developers to 
 prototype new scheduler features and predict their behavior and performance 
 with reasonable amount of confidence, there-by aiding rapid innovation.
 The simulator will exercise the real Yarn ResourceManager removing the 
 network factor by simulating NodeManagers and ApplicationMasters via handling 
 and dispatching NM/AMs heartbeat events from within the same JVM.
 To keep tracking of scheduler behavior and performance, a scheduler wrapper 
 will wrap the real scheduler.
 The simulator will produce real time metrics while executing, including:
 * Resource usages for whole cluster and each queue, which can be utilized to 
 configure cluster and queue's capacity.
 * The detailed application execution trace (recorded in relation to simulated 
 time), which can be analyzed to understand/validate the  scheduler behavior 
 (individual jobs turn around time, throughput, fairness, capacity guarantee, 
 etc).
 * Several key metrics of scheduler algorithm, such as time cost of each 
 scheduler operation (allocate, handle, etc), which can be utilized by Hadoop 
 developers to find the code spots and scalability limits.
 The simulator will provide real time charts showing the behavior of the 
 scheduler and its performance.
 A short demo is available http://www.youtube.com/watch?v=6thLi8q0qLE, showing 
 how to use simulator to simulate Fair Scheduler and Capacity Scheduler.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira


[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2013-08-06 Thread Hadoop QA (JIRA)

[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13731474#comment-13731474
 ] 

Hadoop QA commented on YARN-1021:
-

{color:red}-1 overall{color}.  Here are the results of testing the latest 
attachment 
  http://issues.apache.org/jira/secure/attachment/12596449/YARN-1021.patch
  against trunk revision .

{color:green}+1 @author{color}.  The patch does not contain any @author 
tags.

{color:green}+1 tests included{color}.  The patch appears to include 4 new 
or modified test files.

  {color:red}-1 javac{color}.  The applied patch generated 1163 javac 
compiler warnings (more than the trunk's current 1147 warnings).

{color:green}+1 javadoc{color}.  The javadoc tool did not generate any 
warning messages.

{color:green}+1 eclipse:eclipse{color}.  The patch built with 
eclipse:eclipse.

{color:red}-1 findbugs{color}.  The patch appears to introduce 28 new 
Findbugs (version 1.3.9) warnings.

{color:red}-1 release audit{color}.  The applied patch generated 7 
release audit warnings.

{color:green}+1 core tests{color}.  The patch passed unit tests in 
hadoop-assemblies hadoop-tools/hadoop-sls hadoop-tools/hadoop-tools-dist.

{color:green}+1 contrib tests{color}.  The patch passed contrib unit tests.

Test results: 
https://builds.apache.org/job/PreCommit-YARN-Build/1663//testReport/
Release audit warnings: 
https://builds.apache.org/job/PreCommit-YARN-Build/1663//artifact/trunk/patchprocess/patchReleaseAuditProblems.txt
Findbugs warnings: 
https://builds.apache.org/job/PreCommit-YARN-Build/1663//artifact/trunk/patchprocess/newPatchFindbugsWarningshadoop-sls.html
Javac warnings: 
https://builds.apache.org/job/PreCommit-YARN-Build/1663//artifact/trunk/patchprocess/diffJavacWarnings.txt
Console output: https://builds.apache.org/job/PreCommit-YARN-Build/1663//console

This message is automatically generated.

 Yarn Scheduler Load Simulator
 -

 Key: YARN-1021
 URL: https://issues.apache.org/jira/browse/YARN-1021
 Project: Hadoop YARN
  Issue Type: New Feature
  Components: scheduler
Reporter: Wei Yan
Assignee: Wei Yan
 Attachments: YARN-1021-demo.tar.gz, YARN-1021-images.tar.gz, 
 YARN-1021.patch, YARN-1021.pdf


 The Yarn Scheduler is a fertile area of interest with different 
 implementations, e.g., Fifo, Capacity and Fair  schedulers. Meanwhile, 
 several optimizations are also made to improve scheduler performance for 
 different scenarios and workload. Each scheduler algorithm has its own set of 
 features, and drives scheduling decisions by many factors, such as fairness, 
 capacity guarantee, resource availability, etc. It is very important to 
 evaluate a scheduler algorithm very well before we deploy it in a production 
 cluster. Unfortunately, currently it is non-trivial to evaluate a scheduling 
 algorithm. Evaluating in a real cluster is always time and cost consuming, 
 and it is also very hard to find a large-enough cluster. Hence, a simulator 
 which can predict how well a scheduler algorithm for some specific workload 
 would be quite useful.
 We want to build a Scheduler Load Simulator to simulate large-scale Yarn 
 clusters and application loads in a single machine. This would be invaluable 
 in furthering Yarn by providing a tool for researchers and developers to 
 prototype new scheduler features and predict their behavior and performance 
 with reasonable amount of confidence, there-by aiding rapid innovation.
 The simulator will exercise the real Yarn ResourceManager removing the 
 network factor by simulating NodeManagers and ApplicationMasters via handling 
 and dispatching NM/AMs heartbeat events from within the same JVM.
 To keep tracking of scheduler behavior and performance, a scheduler wrapper 
 will wrap the real scheduler.
 The simulator will produce real time metrics while executing, including:
 * Resource usages for whole cluster and each queue, which can be utilized to 
 configure cluster and queue's capacity.
 * The detailed application execution trace (recorded in relation to simulated 
 time), which can be analyzed to understand/validate the  scheduler behavior 
 (individual jobs turn around time, throughput, fairness, capacity guarantee, 
 etc).
 * Several key metrics of scheduler algorithm, such as time cost of each 
 scheduler operation (allocate, handle, etc), which can be utilized by Hadoop 
 developers to find the code spots and scalability limits.
 The simulator will provide real time charts showing the behavior of the 
 scheduler and its performance.
 A short demo is available http://www.youtube.com/watch?v=6thLi8q0qLE, showing 
 how to use simulator to simulate Fair Scheduler and Capacity Scheduler.

--
This message is automatically generated by JIRA.
If you think it was 

[jira] [Commented] (YARN-1021) Yarn Scheduler Load Simulator

2013-08-05 Thread Bikas Saha (JIRA)

[ 
https://issues.apache.org/jira/browse/YARN-1021?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13730316#comment-13730316
 ] 

Bikas Saha commented on YARN-1021:
--

The idea and goals are very interesting. It would be great if there was a 
design description to initiate a discussion.

 Yarn Scheduler Load Simulator
 -

 Key: YARN-1021
 URL: https://issues.apache.org/jira/browse/YARN-1021
 Project: Hadoop YARN
  Issue Type: New Feature
  Components: scheduler
Reporter: Wei Yan
Assignee: Wei Yan

 The Yarn Scheduler is a fertile area of interest with different 
 implementations, e.g., Fifo, Capacity and Fair  schedulers. Meanwhile, 
 several optimizations are also made to improve scheduler performance for 
 different scenarios and workload. Each scheduler algorithm has its own set of 
 features, and drives scheduling decisions by many factors, such as fairness, 
 capacity guarantee, resource availability, etc. It is very important to 
 evaluate a scheduler algorithm very well before we deploy it in a production 
 cluster. Unfortunately, currently it is non-trivial to evaluate a scheduling 
 algorithm. Evaluating in a real cluster is always time and cost consuming, 
 and it is also very hard to find a large-enough cluster. Hence, a simulator 
 which can predict how well a scheduler algorithm for some specific workload 
 would be quite useful.
 We want to build a Scheduler Load Simulator to simulate large-scale Yarn 
 clusters and application loads in a single machine. This would be invaluable 
 in furthering Yarn by providing a tool for researchers and developers to 
 prototype new scheduler features and predict their behavior and performance 
 with reasonable amount of confidence, there-by aiding rapid innovation.
 The simulator will exercise the real Yarn ResourceManager removing the 
 network factor by simulating NodeManagers and ApplicationMasters via handling 
 and dispatching NM/AMs heartbeat events from within the same JVM.
 To keep tracking of scheduler behavior and performance, a scheduler wrapper 
 will wrap the real scheduler.
 The simulator will produce real time metrics while executing, including:
 * Resource usages for whole cluster and each queue, which can be utilized to 
 configure cluster and queue's capacity.
 * The detailed application execution trace (recorded in relation to simulated 
 time), which can be analyzed to understand/validate the  scheduler behavior 
 (individual jobs turn around time, throughput, fairness, capacity guarantee, 
 etc).
 * Several key metrics of scheduler algorithm, such as time cost of each 
 scheduler operation (allocate, handle, etc), which can be utilized by Hadoop 
 developers to find the code spots and scalability limits.
 The simulator will provide real time charts showing the behavior of the 
 scheduler and its performance.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira