You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@hbase.apache.org by "Todd Lipcon (JIRA)" <ji...@apache.org> on 2011/08/30 03:28:38 UTC
[jira] [Created] (HBASE-4294) HLogSplitter sleeps with 1-second
granularity
HLogSplitter sleeps with 1-second granularity
---------------------------------------------
Key: HBASE-4294
URL: https://issues.apache.org/jira/browse/HBASE-4294
Project: HBase
Issue Type: Bug
Components: master
Affects Versions: 0.90.4
Reporter: Todd Lipcon
Assignee: Todd Lipcon
Priority: Trivial
Fix For: 0.90.5
The HLogSplitter writer threads wait in 1-second intervals for data. When data arrives, notify() is called and they wake up. But when the main thread needs to notify the writers to shut down, it doesn't notify. Hence, the writers always wait up to a full extra second for each log split.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Updated] (HBASE-4294) HLogSplitter sleeps with 1-second
granularity
Posted by "Todd Lipcon (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HBASE-4294?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Todd Lipcon updated HBASE-4294:
-------------------------------
Attachment: 0009-HBASE-4294.-Fix-1-second-sleep-in-HLogSplitter.patch
Trivial patch, verified on cluster that my log split time went from ~2000ms to 1300ms.
> HLogSplitter sleeps with 1-second granularity
> ---------------------------------------------
>
> Key: HBASE-4294
> URL: https://issues.apache.org/jira/browse/HBASE-4294
> Project: HBase
> Issue Type: Bug
> Components: master
> Affects Versions: 0.90.4
> Reporter: Todd Lipcon
> Assignee: Todd Lipcon
> Priority: Trivial
> Fix For: 0.90.5
>
> Attachments: 0009-HBASE-4294.-Fix-1-second-sleep-in-HLogSplitter.patch
>
>
> The HLogSplitter writer threads wait in 1-second intervals for data. When data arrives, notify() is called and they wake up. But when the main thread needs to notify the writers to shut down, it doesn't notify. Hence, the writers always wait up to a full extra second for each log split.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (HBASE-4294) HLogSplitter sleeps with 1-second
granularity
Posted by "Hudson (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HBASE-4294?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13094191#comment-13094191 ]
Hudson commented on HBASE-4294:
-------------------------------
Integrated in HBase-TRUNK #2163 (See [https://builds.apache.org/job/HBase-TRUNK/2163/])
HBASE-4294 HLogSplitter sleeps with 1-second granularity
todd :
Files :
* /hbase/trunk/CHANGES.txt
* /hbase/trunk/src/main/java/org/apache/hadoop/hbase/regionserver/wal/HLogSplitter.java
> HLogSplitter sleeps with 1-second granularity
> ---------------------------------------------
>
> Key: HBASE-4294
> URL: https://issues.apache.org/jira/browse/HBASE-4294
> Project: HBase
> Issue Type: Bug
> Components: master
> Affects Versions: 0.90.4
> Reporter: Todd Lipcon
> Assignee: Todd Lipcon
> Priority: Trivial
> Fix For: 0.90.5
>
> Attachments: 0009-HBASE-4294.-Fix-1-second-sleep-in-HLogSplitter.patch
>
>
> The HLogSplitter writer threads wait in 1-second intervals for data. When data arrives, notify() is called and they wake up. But when the main thread needs to notify the writers to shut down, it doesn't notify. Hence, the writers always wait up to a full extra second for each log split.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (HBASE-4294) HLogSplitter sleeps with 1-second
granularity
Posted by "stack (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HBASE-4294?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13093412#comment-13093412 ]
stack commented on HBASE-4294:
------------------------------
+1
> HLogSplitter sleeps with 1-second granularity
> ---------------------------------------------
>
> Key: HBASE-4294
> URL: https://issues.apache.org/jira/browse/HBASE-4294
> Project: HBase
> Issue Type: Bug
> Components: master
> Affects Versions: 0.90.4
> Reporter: Todd Lipcon
> Assignee: Todd Lipcon
> Priority: Trivial
> Fix For: 0.90.5
>
> Attachments: 0009-HBASE-4294.-Fix-1-second-sleep-in-HLogSplitter.patch
>
>
> The HLogSplitter writer threads wait in 1-second intervals for data. When data arrives, notify() is called and they wake up. But when the main thread needs to notify the writers to shut down, it doesn't notify. Hence, the writers always wait up to a full extra second for each log split.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Resolved] (HBASE-4294) HLogSplitter sleeps with 1-second
granularity
Posted by "Todd Lipcon (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HBASE-4294?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Todd Lipcon resolved HBASE-4294.
--------------------------------
Resolution: Fixed
Hadoop Flags: [Reviewed]
Committed to trunk and branch. I ran TestHLogSplit under jcarder to make sure there weren't any lock inversions here
> HLogSplitter sleeps with 1-second granularity
> ---------------------------------------------
>
> Key: HBASE-4294
> URL: https://issues.apache.org/jira/browse/HBASE-4294
> Project: HBase
> Issue Type: Bug
> Components: master
> Affects Versions: 0.90.4
> Reporter: Todd Lipcon
> Assignee: Todd Lipcon
> Priority: Trivial
> Fix For: 0.90.5
>
> Attachments: 0009-HBASE-4294.-Fix-1-second-sleep-in-HLogSplitter.patch
>
>
> The HLogSplitter writer threads wait in 1-second intervals for data. When data arrives, notify() is called and they wake up. But when the main thread needs to notify the writers to shut down, it doesn't notify. Hence, the writers always wait up to a full extra second for each log split.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira