[
https://issues.apache.org/jira/browse/TEZ-2914?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16226063#comment-16226063
]
Prasanth Jayachandran commented on TEZ-2914:
[~aplusplus] Thanks for the pointer! This will
[
https://issues.apache.org/jira/browse/TEZ-2914?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16224130#comment-16224130
]
Zhiyuan Yang commented on TEZ-2914:
---
CC [~prasanth_j] in case you need this in Hive workload management
>
[
https://issues.apache.org/jira/browse/TEZ-2914?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15071789#comment-15071789
]
TezQA commented on TEZ-2914:
{color:red}-1 overall{color}. Here are the results of testing the latest
[
https://issues.apache.org/jira/browse/TEZ-2914?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15071814#comment-15071814
]
TezQA commented on TEZ-2914:
{color:green}+1 overall{color}. Here are the results of testing the latest
[
https://issues.apache.org/jira/browse/TEZ-2914?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15071811#comment-15071811
]
Bikas Saha commented on TEZ-2914:
-
.4 patch fixes the failing test. .3 patch was stale. Committing.
>
[
https://issues.apache.org/jira/browse/TEZ-2914?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15071770#comment-15071770
]
TezQA commented on TEZ-2914:
{color:red}-1 overall{color}. Here are the results of testing the latest
[
https://issues.apache.org/jira/browse/TEZ-2914?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15070560#comment-15070560
]
Bikas Saha commented on TEZ-2914:
-
Do you mean leave vertexInfo null and do null check instead of get()? If
[
https://issues.apache.org/jira/browse/TEZ-2914?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15070640#comment-15070640
]
Rajesh Balamohan commented on TEZ-2914:
---
Yes, that would avoid the map get call in normal path. Rest
[
https://issues.apache.org/jira/browse/TEZ-2914?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15069370#comment-15069370
]
Rajesh Balamohan commented on TEZ-2914:
---
In most cases, concurrency would be set to -1 (as in
[
https://issues.apache.org/jira/browse/TEZ-2914?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15062893#comment-15062893
]
Hitesh Shah commented on TEZ-2914:
--
Sorry - missed this. Will try and review shortly.
Noticed a bug:
[
https://issues.apache.org/jira/browse/TEZ-2914?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15062900#comment-15062900
]
Hitesh Shah commented on TEZ-2914:
--
Another question related to speculation - should speculation be allowed
[
https://issues.apache.org/jira/browse/TEZ-2914?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15062864#comment-15062864
]
Bikas Saha commented on TEZ-2914:
-
Getting a review on this would be great!
> Ability to limit vertex
[
https://issues.apache.org/jira/browse/TEZ-2914?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15063318#comment-15063318
]
Bikas Saha commented on TEZ-2914:
-
bq. if (vInfo.concurrency >= vInfo.concurrencyLimit) {
We will come here
[
https://issues.apache.org/jira/browse/TEZ-2914?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15056968#comment-15056968
]
Bikas Saha commented on TEZ-2914:
-
[~hitesh] please review!
> Ability to limit vertex concurrency
>
[
https://issues.apache.org/jira/browse/TEZ-2914?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15037173#comment-15037173
]
Bikas Saha commented on TEZ-2914:
-
Attaching patch with the DAGScheduler approach with a unit and system
[
https://issues.apache.org/jira/browse/TEZ-2914?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15034642#comment-15034642
]
Jason Lowe commented on TEZ-2914:
-
One advantage to doing it at the YARN level is that we can tell the RM
[
https://issues.apache.org/jira/browse/TEZ-2914?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15034730#comment-15034730
]
Bikas Saha commented on TEZ-2914:
-
We use AMRMClient for all the RM negotiations. At that level, we cannot
[
https://issues.apache.org/jira/browse/TEZ-2914?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15034325#comment-15034325
]
Bikas Saha commented on TEZ-2914:
-
At least 2 ways to do this.
1) Make the change in the yarn task scheduler
[
https://issues.apache.org/jira/browse/TEZ-2914?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=14978992#comment-14978992
]
Hitesh Shah commented on TEZ-2914:
--
Is this meant to be a global cap for the AM or a per vertex?
>
[
https://issues.apache.org/jira/browse/TEZ-2914?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=14979007#comment-14979007
]
Jonathan Eagles commented on TEZ-2914:
--
[~rohini], [~jlowe], can you help to clarify requirements for
[
https://issues.apache.org/jira/browse/TEZ-2914?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=14979050#comment-14979050
]
Jason Lowe commented on TEZ-2914:
-
If we're trying to port the capability of MAPREDUCE-5583 then it would be
[
https://issues.apache.org/jira/browse/TEZ-2914?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=14979760#comment-14979760
]
Hitesh Shah commented on TEZ-2914:
--
bq. Even with capacity scheduler and hierarchical queues, with the
[
https://issues.apache.org/jira/browse/TEZ-2914?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=14979285#comment-14979285
]
Hitesh Shah commented on TEZ-2914:
--
\cc [~bikassaha] [~sseth]
If the intention is to ensure good job
[
https://issues.apache.org/jira/browse/TEZ-2914?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=14979622#comment-14979622
]
Rohini Palaniswamy commented on TEZ-2914:
-
bq. a cluster resource cap might be something to think
[
https://issues.apache.org/jira/browse/TEZ-2914?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=14979626#comment-14979626
]
Rohini Palaniswamy commented on TEZ-2914:
-
bq. I'm assuming it will simply apply the map property to
25 matches
Mail list logo