[jira] [Commented] (SPARK-25380) Generated plans occupy over 50% of Spark driver memory

2019-06-05 Thread Iris Shaibsky (JIRA)


[ 
https://issues.apache.org/jira/browse/SPARK-25380?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16856470#comment-16856470
 ] 

Iris Shaibsky commented on SPARK-25380:
---

We are facing it also on spark 2.4.2, I see that the PR is merged to master on 
March 13 , but was not included in spark 2.4.3 release. 

When this PR will be included in a release? 

Thanks!

> Generated plans occupy over 50% of Spark driver memory
> --
>
> Key: SPARK-25380
> URL: https://issues.apache.org/jira/browse/SPARK-25380
> Project: Spark
>  Issue Type: Bug
>  Components: Spark Core
>Affects Versions: 2.3.1
> Environment: Spark 2.3.1 (AWS emr-5.16.0)
>  
>Reporter: Michael Spector
>Priority: Minor
> Attachments: Screen Shot 2018-09-06 at 23.19.56.png, Screen Shot 
> 2018-09-12 at 8.20.05.png, heapdump_OOM.png, image-2018-09-16-14-21-38-939.png
>
>
> When debugging an OOM exception during long run of a Spark application (many 
> iterations of the same code) I've found that generated plans occupy most of 
> the driver memory. I'm not sure whether this is a memory leak or not, but it 
> would be helpful if old plans could be purged from memory anyways.
> Attached are screenshots of OOM heap dump opened in JVisualVM.
>  



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

-
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org



[jira] [Commented] (SPARK-25380) Generated plans occupy over 50% of Spark driver memory

2019-02-15 Thread t oo (JIRA)


[ 
https://issues.apache.org/jira/browse/SPARK-25380?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16769724#comment-16769724
 ] 

t oo commented on SPARK-25380:
--

i am also facing issue on spark2.3

> Generated plans occupy over 50% of Spark driver memory
> --
>
> Key: SPARK-25380
> URL: https://issues.apache.org/jira/browse/SPARK-25380
> Project: Spark
>  Issue Type: Bug
>  Components: Spark Core
>Affects Versions: 2.3.1
> Environment: Spark 2.3.1 (AWS emr-5.16.0)
>  
>Reporter: Michael Spector
>Priority: Minor
> Attachments: Screen Shot 2018-09-06 at 23.19.56.png, Screen Shot 
> 2018-09-12 at 8.20.05.png, heapdump_OOM.png, image-2018-09-16-14-21-38-939.png
>
>
> When debugging an OOM exception during long run of a Spark application (many 
> iterations of the same code) I've found that generated plans occupy most of 
> the driver memory. I'm not sure whether this is a memory leak or not, but it 
> would be helpful if old plans could be purged from memory anyways.
> Attached are screenshots of OOM heap dump opened in JVisualVM.
>  



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

-
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org



[jira] [Commented] (SPARK-25380) Generated plans occupy over 50% of Spark driver memory

2018-11-28 Thread Dave DeCaprio (JIRA)


[ 
https://issues.apache.org/jira/browse/SPARK-25380?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16702362#comment-16702362
 ] 

Dave DeCaprio commented on SPARK-25380:
---

I've created a PR that should address this.  It limits the size of text plans 
that are created.  [https://github.com/apache/spark/pull/23169] 

 

> Generated plans occupy over 50% of Spark driver memory
> --
>
> Key: SPARK-25380
> URL: https://issues.apache.org/jira/browse/SPARK-25380
> Project: Spark
>  Issue Type: Bug
>  Components: Spark Core
>Affects Versions: 2.3.1
> Environment: Spark 2.3.1 (AWS emr-5.16.0)
>  
>Reporter: Michael Spector
>Priority: Minor
> Attachments: Screen Shot 2018-09-06 at 23.19.56.png, Screen Shot 
> 2018-09-12 at 8.20.05.png, heapdump_OOM.png, image-2018-09-16-14-21-38-939.png
>
>
> When debugging an OOM exception during long run of a Spark application (many 
> iterations of the same code) I've found that generated plans occupy most of 
> the driver memory. I'm not sure whether this is a memory leak or not, but it 
> would be helpful if old plans could be purged from memory anyways.
> Attached are screenshots of OOM heap dump opened in JVisualVM.
>  



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

-
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org



[jira] [Commented] (SPARK-25380) Generated plans occupy over 50% of Spark driver memory

2018-11-19 Thread Dave DeCaprio (JIRA)


[ 
https://issues.apache.org/jira/browse/SPARK-25380?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16691519#comment-16691519
 ] 

Dave DeCaprio commented on SPARK-25380:
---

I would like to comment that we are also seeing this.  200Mb plans are not 
unusual for us.

> Generated plans occupy over 50% of Spark driver memory
> --
>
> Key: SPARK-25380
> URL: https://issues.apache.org/jira/browse/SPARK-25380
> Project: Spark
>  Issue Type: Bug
>  Components: Spark Core
>Affects Versions: 2.3.1
> Environment: Spark 2.3.1 (AWS emr-5.16.0)
>  
>Reporter: Michael Spector
>Priority: Minor
> Attachments: Screen Shot 2018-09-06 at 23.19.56.png, Screen Shot 
> 2018-09-12 at 8.20.05.png, heapdump_OOM.png, image-2018-09-16-14-21-38-939.png
>
>
> When debugging an OOM exception during long run of a Spark application (many 
> iterations of the same code) I've found that generated plans occupy most of 
> the driver memory. I'm not sure whether this is a memory leak or not, but it 
> would be helpful if old plans could be purged from memory anyways.
> Attached are screenshots of OOM heap dump opened in JVisualVM.
>  



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

-
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org



[jira] [Commented] (SPARK-25380) Generated plans occupy over 50% of Spark driver memory

2018-09-29 Thread Marcelo Vanzin (JIRA)


[ 
https://issues.apache.org/jira/browse/SPARK-25380?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16633145#comment-16633145
 ] 

Marcelo Vanzin commented on SPARK-25380:


We can provide ways to diminish the effect of large plans on memory usage even 
if we can't reproduce his specific case. All the things you list on your last 
e-mail do not need a reproduction; you can hack the code to generate a large 
garbage plan, and you should be able to test any of those solutions.

It would be great to know more and know whether we can make the plans more 
compact; but we should also realize that people can and do run very large and 
complicated queries that generate large plans, and we could help them with 
tuning their UI so not use so much memory.

> Generated plans occupy over 50% of Spark driver memory
> --
>
> Key: SPARK-25380
> URL: https://issues.apache.org/jira/browse/SPARK-25380
> Project: Spark
>  Issue Type: Bug
>  Components: Spark Core
>Affects Versions: 2.3.1
> Environment: Spark 2.3.1 (AWS emr-5.16.0)
>  
>Reporter: Michael Spector
>Priority: Minor
> Attachments: Screen Shot 2018-09-06 at 23.19.56.png, Screen Shot 
> 2018-09-12 at 8.20.05.png, heapdump_OOM.png, image-2018-09-16-14-21-38-939.png
>
>
> When debugging an OOM exception during long run of a Spark application (many 
> iterations of the same code) I've found that generated plans occupy most of 
> the driver memory. I'm not sure whether this is a memory leak or not, but it 
> would be helpful if old plans could be purged from memory anyways.
> Attached are screenshots of OOM heap dump opened in JVisualVM.
>  



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

-
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org



[jira] [Commented] (SPARK-25380) Generated plans occupy over 50% of Spark driver memory

2018-09-27 Thread Jungtaek Lim (JIRA)


[ 
https://issues.apache.org/jira/browse/SPARK-25380?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16631313#comment-16631313
 ] 

Jungtaek Lim commented on SPARK-25380:
--

Btw, reproducer still helps when we tackle it with only UI side. There're a few 
options to avoid memory issue:
 # Remove feature (or have option to "opt out") on showing physical plan 
description in UI.
 # Find some ways to dramatically reduce memory on storing physical plan 
description.
 # Purge old physical plan (count or memory based).

2 and 3 can be applied individually or together. If we are interested on 2, we 
would still want to have actual string to see how we reduce it (like 
compression/decompression).

> Generated plans occupy over 50% of Spark driver memory
> --
>
> Key: SPARK-25380
> URL: https://issues.apache.org/jira/browse/SPARK-25380
> Project: Spark
>  Issue Type: Bug
>  Components: Spark Core
>Affects Versions: 2.3.1
> Environment: Spark 2.3.1 (AWS emr-5.16.0)
>  
>Reporter: Michael Spector
>Priority: Minor
> Attachments: Screen Shot 2018-09-06 at 23.19.56.png, Screen Shot 
> 2018-09-12 at 8.20.05.png, heapdump_OOM.png, image-2018-09-16-14-21-38-939.png
>
>
> When debugging an OOM exception during long run of a Spark application (many 
> iterations of the same code) I've found that generated plans occupy most of 
> the driver memory. I'm not sure whether this is a memory leak or not, but it 
> would be helpful if old plans could be purged from memory anyways.
> Attached are screenshots of OOM heap dump opened in JVisualVM.
>  



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

-
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org



[jira] [Commented] (SPARK-25380) Generated plans occupy over 50% of Spark driver memory

2018-09-27 Thread Jungtaek Lim (JIRA)


[ 
https://issues.apache.org/jira/browse/SPARK-25380?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16631295#comment-16631295
 ] 

Jungtaek Lim commented on SPARK-25380:
--

IMHO it depends on how we see the issue and how we would like to tackle this.

If we think 200M of plan string is normal and usual, you're right the issue 
lays in UI and UI should deal with it well.
(Even 200M of single plan would be out of expectation on end users and they 
might miss to consider allocating enough space on driver side for UI, so 
purging old plan would work for some cases but not for some other cases.)

If we don't think 200M of plan string is normal, we need to see actual case and 
investigate which physical node occupies much space on representing, and 
whether they're really needed or too verbose. If the huge string came from 
representing physical node itself which doesn't change among batches, we may be 
able to try storing template format of message for physical node and variables 
separately and apply just when page is requested.

If we know more, we could have better solution: according to your previous 
comment, I guess we're on the same page:
{quote}They seem to hold a lot more memory than just the plan graph structures 
do, it would be nice to know what exactly is holding on to that memory.
{quote}
Since we are unlikely to get reproducer, I wouldn't want to block anyone to 
work on this. Anyone could tackle on UI issue.

> Generated plans occupy over 50% of Spark driver memory
> --
>
> Key: SPARK-25380
> URL: https://issues.apache.org/jira/browse/SPARK-25380
> Project: Spark
>  Issue Type: Bug
>  Components: Spark Core
>Affects Versions: 2.3.1
> Environment: Spark 2.3.1 (AWS emr-5.16.0)
>  
>Reporter: Michael Spector
>Priority: Minor
> Attachments: Screen Shot 2018-09-06 at 23.19.56.png, Screen Shot 
> 2018-09-12 at 8.20.05.png, heapdump_OOM.png, image-2018-09-16-14-21-38-939.png
>
>
> When debugging an OOM exception during long run of a Spark application (many 
> iterations of the same code) I've found that generated plans occupy most of 
> the driver memory. I'm not sure whether this is a memory leak or not, but it 
> would be helpful if old plans could be purged from memory anyways.
> Attached are screenshots of OOM heap dump opened in JVisualVM.
>  



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

-
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org



[jira] [Commented] (SPARK-25380) Generated plans occupy over 50% of Spark driver memory

2018-09-27 Thread Marcelo Vanzin (JIRA)


[ 
https://issues.apache.org/jira/browse/SPARK-25380?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16631236#comment-16631236
 ] 

Marcelo Vanzin commented on SPARK-25380:


If all you want is see this live, why do you need a query that generates a 
large plan?

Just hack whatever method creates the plan string to return a really large 
string full of garbage. Same result.

The goal here is not to reproduce the original problem, but to provide a 
solution for when that happens; there's no problem with his query other than 
the plan being large and the UI not dealing with that well.

> Generated plans occupy over 50% of Spark driver memory
> --
>
> Key: SPARK-25380
> URL: https://issues.apache.org/jira/browse/SPARK-25380
> Project: Spark
>  Issue Type: Bug
>  Components: Spark Core
>Affects Versions: 2.3.1
> Environment: Spark 2.3.1 (AWS emr-5.16.0)
>  
>Reporter: Michael Spector
>Priority: Minor
> Attachments: Screen Shot 2018-09-06 at 23.19.56.png, Screen Shot 
> 2018-09-12 at 8.20.05.png, heapdump_OOM.png, image-2018-09-16-14-21-38-939.png
>
>
> When debugging an OOM exception during long run of a Spark application (many 
> iterations of the same code) I've found that generated plans occupy most of 
> the driver memory. I'm not sure whether this is a memory leak or not, but it 
> would be helpful if old plans could be purged from memory anyways.
> Attached are screenshots of OOM heap dump opened in JVisualVM.
>  



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

-
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org



[jira] [Commented] (SPARK-25380) Generated plans occupy over 50% of Spark driver memory

2018-09-27 Thread Jungtaek Lim (JIRA)


[ 
https://issues.apache.org/jira/browse/SPARK-25380?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16630053#comment-16630053
 ] 

Jungtaek Lim commented on SPARK-25380:
--

[~spektom]

I guess physical plan description being 200M is unusual case, so I would spent 
most of efforts finding how to reproduce which I wouldn't want to. If you 
couldn't share the reproducer (or the way how to make them bigger) I would just 
give up.

> Generated plans occupy over 50% of Spark driver memory
> --
>
> Key: SPARK-25380
> URL: https://issues.apache.org/jira/browse/SPARK-25380
> Project: Spark
>  Issue Type: Bug
>  Components: Spark Core
>Affects Versions: 2.3.1
> Environment: Spark 2.3.1 (AWS emr-5.16.0)
>  
>Reporter: Michael Spector
>Priority: Minor
> Attachments: Screen Shot 2018-09-06 at 23.19.56.png, Screen Shot 
> 2018-09-12 at 8.20.05.png, heapdump_OOM.png, image-2018-09-16-14-21-38-939.png
>
>
> When debugging an OOM exception during long run of a Spark application (many 
> iterations of the same code) I've found that generated plans occupy most of 
> the driver memory. I'm not sure whether this is a memory leak or not, but it 
> would be helpful if old plans could be purged from memory anyways.
> Attached are screenshots of OOM heap dump opened in JVisualVM.
>  



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

-
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org



[jira] [Commented] (SPARK-25380) Generated plans occupy over 50% of Spark driver memory

2018-09-24 Thread Michael Spector (JIRA)


[ 
https://issues.apache.org/jira/browse/SPARK-25380?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16626790#comment-16626790
 ] 

Michael Spector commented on SPARK-25380:
-

It's really difficult to extract a working reproduction case from the
application we have. What would help you for understanding the problem
theoretically?




> Generated plans occupy over 50% of Spark driver memory
> --
>
> Key: SPARK-25380
> URL: https://issues.apache.org/jira/browse/SPARK-25380
> Project: Spark
>  Issue Type: Bug
>  Components: Spark Core
>Affects Versions: 2.3.1
> Environment: Spark 2.3.1 (AWS emr-5.16.0)
>  
>Reporter: Michael Spector
>Priority: Minor
> Attachments: Screen Shot 2018-09-06 at 23.19.56.png, Screen Shot 
> 2018-09-12 at 8.20.05.png, heapdump_OOM.png, image-2018-09-16-14-21-38-939.png
>
>
> When debugging an OOM exception during long run of a Spark application (many 
> iterations of the same code) I've found that generated plans occupy most of 
> the driver memory. I'm not sure whether this is a memory leak or not, but it 
> would be helpful if old plans could be purged from memory anyways.
> Attached are screenshots of OOM heap dump opened in JVisualVM.
>  



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

-
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org



[jira] [Commented] (SPARK-25380) Generated plans occupy over 50% of Spark driver memory

2018-09-20 Thread Jungtaek Lim (JIRA)


[ 
https://issues.apache.org/jira/browse/SPARK-25380?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16623087#comment-16623087
 ] 

Jungtaek Lim commented on SPARK-25380:
--

I thought about this as edge case which we might be unsure to address in 
general, but once two end users report the same thing it doesn't look like odd 
case.

I'm interested on tackling this issue, but without reproducer I can't play 
with. Could one of you please share "redacted" query which can consistently 
reproduce the issue?

> Generated plans occupy over 50% of Spark driver memory
> --
>
> Key: SPARK-25380
> URL: https://issues.apache.org/jira/browse/SPARK-25380
> Project: Spark
>  Issue Type: Bug
>  Components: Spark Core
>Affects Versions: 2.3.1
> Environment: Spark 2.3.1 (AWS emr-5.16.0)
>  
>Reporter: Michael Spector
>Priority: Minor
> Attachments: Screen Shot 2018-09-06 at 23.19.56.png, Screen Shot 
> 2018-09-12 at 8.20.05.png, heapdump_OOM.png, image-2018-09-16-14-21-38-939.png
>
>
> When debugging an OOM exception during long run of a Spark application (many 
> iterations of the same code) I've found that generated plans occupy most of 
> the driver memory. I'm not sure whether this is a memory leak or not, but it 
> would be helpful if old plans could be purged from memory anyways.
> Attached are screenshots of OOM heap dump opened in JVisualVM.
>  



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

-
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org



[jira] [Commented] (SPARK-25380) Generated plans occupy over 50% of Spark driver memory

2018-09-16 Thread Nir Hedvat (JIRA)


[ 
https://issues.apache.org/jira/browse/SPARK-25380?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16616675#comment-16616675
 ] 

Nir Hedvat commented on SPARK-25380:


Same problem here (using Spark 2.3.1)

> Generated plans occupy over 50% of Spark driver memory
> --
>
> Key: SPARK-25380
> URL: https://issues.apache.org/jira/browse/SPARK-25380
> Project: Spark
>  Issue Type: Bug
>  Components: Spark Core
>Affects Versions: 2.3.1
> Environment: Spark 2.3.1 (AWS emr-5.16.0)
>  
>Reporter: Michael Spector
>Priority: Minor
> Attachments: Screen Shot 2018-09-06 at 23.19.56.png, Screen Shot 
> 2018-09-12 at 8.20.05.png, heapdump_OOM.png
>
>
> When debugging an OOM exception during long run of a Spark application (many 
> iterations of the same code) I've found that generated plans occupy most of 
> the driver memory. I'm not sure whether this is a memory leak or not, but it 
> would be helpful if old plans could be purged from memory anyways.
> Attached are screenshots of OOM heap dump opened in JVisualVM.
>  



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

-
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org



[jira] [Commented] (SPARK-25380) Generated plans occupy over 50% of Spark driver memory

2018-09-12 Thread Marcelo Vanzin (JIRA)


[ 
https://issues.apache.org/jira/browse/SPARK-25380?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16612433#comment-16612433
 ] 

Marcelo Vanzin commented on SPARK-25380:


Yep. That's a 200MB plan description string...

> Generated plans occupy over 50% of Spark driver memory
> --
>
> Key: SPARK-25380
> URL: https://issues.apache.org/jira/browse/SPARK-25380
> Project: Spark
>  Issue Type: Bug
>  Components: Spark Core
>Affects Versions: 2.3.1
> Environment: Spark 2.3.1 (AWS emr-5.16.0)
>  
>Reporter: Michael Spector
>Priority: Minor
> Attachments: Screen Shot 2018-09-06 at 23.19.56.png, Screen Shot 
> 2018-09-12 at 8.20.05.png, heapdump_OOM.png
>
>
> When debugging an OOM exception during long run of a Spark application (many 
> iterations of the same code) I've found that generated plans occupy most of 
> the driver memory. I'm not sure whether this is a memory leak or not, but it 
> would be helpful if old plans could be purged from memory anyways.
> Attached are screenshots of OOM heap dump opened in JVisualVM.
>  



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

-
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org



[jira] [Commented] (SPARK-25380) Generated plans occupy over 50% of Spark driver memory

2018-09-11 Thread Michael Spector (JIRA)


[ 
https://issues.apache.org/jira/browse/SPARK-25380?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16611606#comment-16611606
 ] 

Michael Spector commented on SPARK-25380:
-

[~vanzin] Here's the breakdown:

!Screen Shot 2018-09-12 at 8.20.05.png|width=1021,height=713!

LMK, if you need more information, I'll be more than glad to help.

> Generated plans occupy over 50% of Spark driver memory
> --
>
> Key: SPARK-25380
> URL: https://issues.apache.org/jira/browse/SPARK-25380
> Project: Spark
>  Issue Type: Bug
>  Components: Spark Core
>Affects Versions: 2.3.1
> Environment: Spark 2.3.1 (AWS emr-5.16.0)
>  
>Reporter: Michael Spector
>Priority: Minor
> Attachments: Screen Shot 2018-09-06 at 23.19.56.png, Screen Shot 
> 2018-09-12 at 8.20.05.png, heapdump_OOM.png
>
>
> When debugging an OOM exception during long run of a Spark application (many 
> iterations of the same code) I've found that generated plans occupy most of 
> the driver memory. I'm not sure whether this is a memory leak or not, but it 
> would be helpful if old plans could be purged from memory anyways.
> Attached are screenshots of OOM heap dump opened in JVisualVM.
>  



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

-
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org



[jira] [Commented] (SPARK-25380) Generated plans occupy over 50% of Spark driver memory

2018-09-11 Thread Marcelo Vanzin (JIRA)


[ 
https://issues.apache.org/jira/browse/SPARK-25380?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16611012#comment-16611012
 ] 

Marcelo Vanzin commented on SPARK-25380:


Another bit of information that would be useful is the breakdown of memory 
usage of the {{SQLExecutionUIData}} instances. They seem to hold a lot more 
memory than just the plan graph structures do, it would be nice to know what 
exactly is holding on to that memory.

(Assuming you can't share the heap dump...)

> Generated plans occupy over 50% of Spark driver memory
> --
>
> Key: SPARK-25380
> URL: https://issues.apache.org/jira/browse/SPARK-25380
> Project: Spark
>  Issue Type: Bug
>  Components: Spark Core
>Affects Versions: 2.3.1
> Environment: Spark 2.3.1 (AWS emr-5.16.0)
>  
>Reporter: Michael Spector
>Priority: Minor
> Attachments: Screen Shot 2018-09-06 at 23.19.56.png, heapdump_OOM.png
>
>
> When debugging an OOM exception during long run of a Spark application (many 
> iterations of the same code) I've found that generated plans occupy most of 
> the driver memory. I'm not sure whether this is a memory leak or not, but it 
> would be helpful if old plans could be purged from memory anyways.
> Attached are screenshots of OOM heap dump opened in JVisualVM.
>  



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

-
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org



[jira] [Commented] (SPARK-25380) Generated plans occupy over 50% of Spark driver memory

2018-09-08 Thread Michael Spector (JIRA)


[ 
https://issues.apache.org/jira/browse/SPARK-25380?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16608045#comment-16608045
 ] 

Michael Spector commented on SPARK-25380:
-

Originally reported in SPARK-23670

 

> Generated plans occupy over 50% of Spark driver memory
> --
>
> Key: SPARK-25380
> URL: https://issues.apache.org/jira/browse/SPARK-25380
> Project: Spark
>  Issue Type: Bug
>  Components: Spark Core
>Affects Versions: 2.3.1
> Environment: Spark 2.3.1 (AWS emr-5.16.0)
>  
>Reporter: Michael Spector
>Priority: Minor
> Attachments: Screen Shot 2018-09-06 at 23.19.56.png, heapdump_OOM.png
>
>
> When debugging an OOM exception during long run of a Spark application (many 
> iterations of the same code) I've found that generated plans occupy most of 
> the driver memory. I'm not sure whether this is a memory leak or not, but it 
> would be helpful if old plans could be purged from memory anyways.
> Attached are screenshots of OOM heap dump opened in JVisualVM.
>  



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

-
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org