You are viewing a plain text version of this content. The canonical link for it is here.
Posted to mapreduce-issues@hadoop.apache.org by "Amol Kekre (Created) (JIRA)" <ji...@apache.org> on 2012/01/06 02:53:39 UTC
[jira] [Created] (MAPREDUCE-3628) DFSIO read throughput is
decreased by 16% in 0.23 than Hadoop-0.20.204 on 350 nodes size cluster.
DFSIO read throughput is decreased by 16% in 0.23 than Hadoop-0.20.204 on 350 nodes size cluster.
-------------------------------------------------------------------------------------------------
Key: MAPREDUCE-3628
URL: https://issues.apache.org/jira/browse/MAPREDUCE-3628
Project: Hadoop Map/Reduce
Issue Type: Task
Components: mrv2
Affects Versions: 0.23.0
Reporter: Amol Kekre
Priority: Critical
Fix For: 0.23.1
DFSIO read throughput is decreased by 16% in 0.23 than Hadoop-0.20.204 on 350 nodes size cluster.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Assigned] (MAPREDUCE-3628) DFSIO read throughput is
decreased by 16% in 0.23 than Hadoop-0.20.204 on 350 nodes size cluster.
Posted by "Mahadev konar (Assigned) (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/MAPREDUCE-3628?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Mahadev konar reassigned MAPREDUCE-3628:
----------------------------------------
Assignee: Vinod Kumar Vavilapalli
> DFSIO read throughput is decreased by 16% in 0.23 than Hadoop-0.20.204 on 350 nodes size cluster.
> -------------------------------------------------------------------------------------------------
>
> Key: MAPREDUCE-3628
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-3628
> Project: Hadoop Map/Reduce
> Issue Type: Task
> Components: mrv2
> Affects Versions: 0.23.0
> Reporter: Amol Kekre
> Assignee: Vinod Kumar Vavilapalli
> Priority: Blocker
> Fix For: 0.23.1
>
>
> DFSIO read throughput is decreased by 16% in 0.23 than Hadoop-0.20.204 on 350 nodes size cluster.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (MAPREDUCE-3628) DFSIO read throughput is
decreased by 16% in 0.23 than Hadoop-0.20.204 on 350 nodes size cluster.
Posted by "Amol Kekre (Commented) (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/MAPREDUCE-3628?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13201547#comment-13201547 ]
Amol Kekre commented on MAPREDUCE-3628:
---------------------------------------
Throughput decrease was 10% with 6% CPU slowdown a few wks ago. Waiting for new numbers later this wk.
> DFSIO read throughput is decreased by 16% in 0.23 than Hadoop-0.20.204 on 350 nodes size cluster.
> -------------------------------------------------------------------------------------------------
>
> Key: MAPREDUCE-3628
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-3628
> Project: Hadoop Map/Reduce
> Issue Type: Sub-task
> Components: mrv2
> Affects Versions: 0.23.0
> Reporter: Amol Kekre
> Assignee: Vinod Kumar Vavilapalli
> Priority: Blocker
> Fix For: 0.23.1
>
>
> DFSIO read throughput is decreased by 16% in 0.23 than Hadoop-0.20.204 on 350 nodes size cluster.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (MAPREDUCE-3628) DFSIO read throughput is
decreased by 16% in 0.23 than Hadoop-0.20.204 on 350 nodes size cluster.
Posted by "Todd Lipcon (Commented) (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/MAPREDUCE-3628?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13181113#comment-13181113 ]
Todd Lipcon commented on MAPREDUCE-3628:
----------------------------------------
Which test case are you using here? Is it TestDFSIO or something closer to a "teravalidate" type workload? Is the input dataset larger than RAM or are you disk-bound?
> DFSIO read throughput is decreased by 16% in 0.23 than Hadoop-0.20.204 on 350 nodes size cluster.
> -------------------------------------------------------------------------------------------------
>
> Key: MAPREDUCE-3628
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-3628
> Project: Hadoop Map/Reduce
> Issue Type: Task
> Components: mrv2
> Affects Versions: 0.23.0
> Reporter: Amol Kekre
> Priority: Critical
> Fix For: 0.23.1
>
>
> DFSIO read throughput is decreased by 16% in 0.23 than Hadoop-0.20.204 on 350 nodes size cluster.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (MAPREDUCE-3628) DFSIO read throughput is
decreased by 16% in 0.23 than Hadoop-0.20.204 on 350 nodes size cluster.
Posted by "Kihwal Lee (Commented) (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/MAPREDUCE-3628?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13181391#comment-13181391 ]
Kihwal Lee commented on MAPREDUCE-3628:
---------------------------------------
I believe it was TestDFSIO.
{noformat}
12/01/05 07:50:01 INFO fs.TestDFSIO: TestDFSIO.0.0.6
12/01/05 07:50:01 INFO fs.TestDFSIO: nrFiles = 2037
12/01/05 07:50:01 INFO fs.TestDFSIO: fileSize (MB) = 1024.0
12/01/05 07:50:01 INFO fs.TestDFSIO: bufferSize = 10485760
12/01/05 07:50:01 INFO fs.TestDFSIO: baseDir = /benchmarks/TestDFSIO
12/01/05 07:50:02 INFO fs.TestDFSIO: creating control file: 1073741824 bytes, 2037 files
12/01/05 07:50:36 INFO fs.TestDFSIO: created control files for: 2037 files
{noformat}
The input data set wasn't that big. So many factors external to DFS and disk I/O could have caused this. I asked Vinay who ran the test for more details.
> DFSIO read throughput is decreased by 16% in 0.23 than Hadoop-0.20.204 on 350 nodes size cluster.
> -------------------------------------------------------------------------------------------------
>
> Key: MAPREDUCE-3628
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-3628
> Project: Hadoop Map/Reduce
> Issue Type: Task
> Components: mrv2
> Affects Versions: 0.23.0
> Reporter: Amol Kekre
> Priority: Blocker
> Fix For: 0.23.1
>
>
> DFSIO read throughput is decreased by 16% in 0.23 than Hadoop-0.20.204 on 350 nodes size cluster.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (MAPREDUCE-3628) DFSIO read throughput is
decreased by 16% in 0.23 than Hadoop-0.20.204 on 350 nodes size cluster.
Posted by "Todd Lipcon (Commented) (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/MAPREDUCE-3628?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13181510#comment-13181510 ]
Todd Lipcon commented on MAPREDUCE-3628:
----------------------------------------
k, my hunch is that it's probably MR-related, then. I've seen a lot of cases where differences in scheduler performance give either better or worse looking DFSIO results on the same HDFS version. And my benchmarks of HDFS 0.23 show it significantly _faster_ than 0.20.
> DFSIO read throughput is decreased by 16% in 0.23 than Hadoop-0.20.204 on 350 nodes size cluster.
> -------------------------------------------------------------------------------------------------
>
> Key: MAPREDUCE-3628
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-3628
> Project: Hadoop Map/Reduce
> Issue Type: Task
> Components: mrv2
> Affects Versions: 0.23.0
> Reporter: Amol Kekre
> Priority: Blocker
> Fix For: 0.23.1
>
>
> DFSIO read throughput is decreased by 16% in 0.23 than Hadoop-0.20.204 on 350 nodes size cluster.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Updated] (MAPREDUCE-3628) DFSIO read throughput is
decreased by 16% in 0.23 than Hadoop-0.20.204 on 350 nodes size cluster.
Posted by "Vinod Kumar Vavilapalli (Updated) (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/MAPREDUCE-3628?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Vinod Kumar Vavilapalli updated MAPREDUCE-3628:
-----------------------------------------------
Issue Type: Sub-task (was: Task)
Parent: MAPREDUCE-3561
> DFSIO read throughput is decreased by 16% in 0.23 than Hadoop-0.20.204 on 350 nodes size cluster.
> -------------------------------------------------------------------------------------------------
>
> Key: MAPREDUCE-3628
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-3628
> Project: Hadoop Map/Reduce
> Issue Type: Sub-task
> Components: mrv2
> Affects Versions: 0.23.0
> Reporter: Amol Kekre
> Assignee: Vinod Kumar Vavilapalli
> Priority: Blocker
> Fix For: 0.23.1
>
>
> DFSIO read throughput is decreased by 16% in 0.23 than Hadoop-0.20.204 on 350 nodes size cluster.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Updated] (MAPREDUCE-3628) DFSIO read throughput is
decreased by 16% in 0.23.1 than Hadoop-0.20.204 on 350 nodes size cluster.
Posted by "Amol Kekre (Updated) (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/MAPREDUCE-3628?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Amol Kekre updated MAPREDUCE-3628:
----------------------------------
Priority: Major (was: Blocker)
Summary: DFSIO read throughput is decreased by 16% in 0.23.1 than Hadoop-0.20.204 on 350 nodes size cluster. (was: DFSIO read throughput is decreased by 16% in 0.23 than Hadoop-0.20.204 on 350 nodes size cluster.)
> DFSIO read throughput is decreased by 16% in 0.23.1 than Hadoop-0.20.204 on 350 nodes size cluster.
> ---------------------------------------------------------------------------------------------------
>
> Key: MAPREDUCE-3628
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-3628
> Project: Hadoop Map/Reduce
> Issue Type: Sub-task
> Components: mrv2
> Affects Versions: 0.23.0
> Reporter: Amol Kekre
> Assignee: Vinod Kumar Vavilapalli
> Fix For: 0.23.1
>
>
> DFSIO read throughput is decreased by 16% in 0.23 than Hadoop-0.20.204 on 350 nodes size cluster.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (MAPREDUCE-3628) DFSIO read throughput is
decreased by 16% in 0.23 than Hadoop-0.20.204 on 350 nodes size cluster.
Posted by "Amol Kekre (Commented) (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/MAPREDUCE-3628?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13211674#comment-13211674 ]
Amol Kekre commented on MAPREDUCE-3628:
---------------------------------------
throughput decrease is now 4%. This jira is no longer a blocker
> DFSIO read throughput is decreased by 16% in 0.23 than Hadoop-0.20.204 on 350 nodes size cluster.
> -------------------------------------------------------------------------------------------------
>
> Key: MAPREDUCE-3628
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-3628
> Project: Hadoop Map/Reduce
> Issue Type: Sub-task
> Components: mrv2
> Affects Versions: 0.23.0
> Reporter: Amol Kekre
> Assignee: Vinod Kumar Vavilapalli
> Priority: Blocker
> Fix For: 0.23.1
>
>
> DFSIO read throughput is decreased by 16% in 0.23 than Hadoop-0.20.204 on 350 nodes size cluster.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (MAPREDUCE-3628) DFSIO read throughput is
decreased by 16% in 0.23 than Hadoop-0.20.204 on 350 nodes size cluster.
Posted by "Amol Kekre (Commented) (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/MAPREDUCE-3628?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13181528#comment-13181528 ]
Amol Kekre commented on MAPREDUCE-3628:
---------------------------------------
Todd, thanks for the update. We are taking a look at our current data.
> DFSIO read throughput is decreased by 16% in 0.23 than Hadoop-0.20.204 on 350 nodes size cluster.
> -------------------------------------------------------------------------------------------------
>
> Key: MAPREDUCE-3628
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-3628
> Project: Hadoop Map/Reduce
> Issue Type: Task
> Components: mrv2
> Affects Versions: 0.23.0
> Reporter: Amol Kekre
> Priority: Blocker
> Fix For: 0.23.1
>
>
> DFSIO read throughput is decreased by 16% in 0.23 than Hadoop-0.20.204 on 350 nodes size cluster.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Resolved] (MAPREDUCE-3628) DFSIO read throughput is
decreased by 16% in 0.23.1 than Hadoop-0.20.204 on 350 nodes size cluster.
Posted by "Vinod Kumar Vavilapalli (Resolved) (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/MAPREDUCE-3628?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Vinod Kumar Vavilapalli resolved MAPREDUCE-3628.
------------------------------------------------
Resolution: Fixed
Fix Version/s: (was: 0.23.1)
Thanks for the update Amol. 4% is fair, we can look at it in the next iteration.
In all likelihood, this got resolved after HDFS-2923. Closing this as fixed.
> DFSIO read throughput is decreased by 16% in 0.23.1 than Hadoop-0.20.204 on 350 nodes size cluster.
> ---------------------------------------------------------------------------------------------------
>
> Key: MAPREDUCE-3628
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-3628
> Project: Hadoop Map/Reduce
> Issue Type: Sub-task
> Components: mrv2
> Affects Versions: 0.23.0
> Reporter: Amol Kekre
> Assignee: Vinod Kumar Vavilapalli
>
> DFSIO read throughput is decreased by 16% in 0.23 than Hadoop-0.20.204 on 350 nodes size cluster.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Updated] (MAPREDUCE-3628) DFSIO read throughput is
decreased by 16% in 0.23 than Hadoop-0.20.204 on 350 nodes size cluster.
Posted by "Arun C Murthy (Updated) (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/MAPREDUCE-3628?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Arun C Murthy updated MAPREDUCE-3628:
-------------------------------------
Priority: Blocker (was: Critical)
> DFSIO read throughput is decreased by 16% in 0.23 than Hadoop-0.20.204 on 350 nodes size cluster.
> -------------------------------------------------------------------------------------------------
>
> Key: MAPREDUCE-3628
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-3628
> Project: Hadoop Map/Reduce
> Issue Type: Task
> Components: mrv2
> Affects Versions: 0.23.0
> Reporter: Amol Kekre
> Priority: Blocker
> Fix For: 0.23.1
>
>
> DFSIO read throughput is decreased by 16% in 0.23 than Hadoop-0.20.204 on 350 nodes size cluster.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira