You are viewing a plain text version of this content. The canonical link for it is here.
Posted to common-dev@hadoop.apache.org by "Runping Qi (JIRA)" <ji...@apache.org> on 2008/04/01 04:44:24 UTC
[jira] Created: (HADOOP-3146) SequenceFile should not call flush
excessively
SequenceFile should not call flush excessively
----------------------------------------------
Key: HADOOP-3146
URL: https://issues.apache.org/jira/browse/HADOOP-3146
Project: Hadoop Core
Issue Type: Bug
Components: mapred
Reporter: Runping Qi
Starting fron hadoop-0.17, calling flush of dfs becomes very expensive.
It does not make much sense to call flush inside append.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Commented: (HADOOP-3146) DFSOutputStream.flush should be
renamed as DFSOutputStream.fsync
Posted by "Hadoop QA (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HADOOP-3146?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12585284#action_12585284 ]
Hadoop QA commented on HADOOP-3146:
-----------------------------------
-1 overall. Here are the results of testing the latest attachment
http://issues.apache.org/jira/secure/attachment/12379196/flushToFsync.patch
against trunk revision 643282.
@author +1. The patch does not contain any @author tags.
tests included +1. The patch appears to include 6 new or modified tests.
javadoc +1. The javadoc tool did not generate any warning messages.
javac +1. The applied patch does not generate any new javac compiler warnings.
release audit +1. The applied patch does not generate any new release audit warnings.
findbugs +1. The patch does not introduce any new Findbugs warnings.
core tests -1. The patch failed core unit tests.
contrib tests +1. The patch passed contrib unit tests.
Test results: http://hudson.zones.apache.org/hudson/job/Hadoop-Patch/2143/testReport/
Findbugs warnings: http://hudson.zones.apache.org/hudson/job/Hadoop-Patch/2143/artifact/trunk/build/test/findbugs/newPatchFindbugsWarnings.html
Checkstyle results: http://hudson.zones.apache.org/hudson/job/Hadoop-Patch/2143/artifact/trunk/build/test/checkstyle-errors.html
Console output: http://hudson.zones.apache.org/hudson/job/Hadoop-Patch/2143/console
This message is automatically generated.
> DFSOutputStream.flush should be renamed as DFSOutputStream.fsync
> ----------------------------------------------------------------
>
> Key: HADOOP-3146
> URL: https://issues.apache.org/jira/browse/HADOOP-3146
> Project: Hadoop Core
> Issue Type: Bug
> Components: dfs
> Reporter: Runping Qi
> Assignee: dhruba borthakur
> Priority: Blocker
> Fix For: 0.17.0
>
> Attachments: flushToFsync.patch
>
>
> Starting fron hadoop-0.17, calling flush of dfs becomes very expensive.
> The current implementation of flush should be renamed as sync.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Commented: (HADOOP-3146) SequenceFile should not call flush
excessively
Posted by "Runping Qi (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HADOOP-3146?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12583995#action_12583995 ]
Runping Qi commented on HADOOP-3146:
------------------------------------
Then the dfs (and maybe sequence file too) need to provide an api similar to sync.
> SequenceFile should not call flush excessively
> ----------------------------------------------
>
> Key: HADOOP-3146
> URL: https://issues.apache.org/jira/browse/HADOOP-3146
> Project: Hadoop Core
> Issue Type: Bug
> Components: dfs
> Reporter: Runping Qi
> Fix For: 0.17.0
>
>
> Starting fron hadoop-0.17, calling flush of dfs becomes very expensive.
> It does not make much sense to call flush inside append.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Commented: (HADOOP-3146) SequenceFile should not call flush
excessively
Posted by "dhruba borthakur (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HADOOP-3146?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12583996#action_12583996 ]
dhruba borthakur commented on HADOOP-3146:
------------------------------------------
I agree. I think we should revert back the semantics of DFSOutputStream.flush() to what it was in 0.16. Also, introduce a new call DFSOutputStream.sync() that actually waits for packets to get acknowledged from datanodes. Patch coming soon....
> SequenceFile should not call flush excessively
> ----------------------------------------------
>
> Key: HADOOP-3146
> URL: https://issues.apache.org/jira/browse/HADOOP-3146
> Project: Hadoop Core
> Issue Type: Bug
> Components: dfs
> Reporter: Runping Qi
> Fix For: 0.17.0
>
>
> Starting fron hadoop-0.17, calling flush of dfs becomes very expensive.
> It does not make much sense to call flush inside append.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Updated: (HADOOP-3146) DFSOutputStream.flush should be
renamed as DFSOutputStream.fsync
Posted by "dhruba borthakur (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HADOOP-3146?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
dhruba borthakur updated HADOOP-3146:
-------------------------------------
Status: Open (was: Patch Available)
> DFSOutputStream.flush should be renamed as DFSOutputStream.fsync
> ----------------------------------------------------------------
>
> Key: HADOOP-3146
> URL: https://issues.apache.org/jira/browse/HADOOP-3146
> Project: Hadoop Core
> Issue Type: Bug
> Components: dfs
> Reporter: Runping Qi
> Assignee: dhruba borthakur
> Priority: Blocker
> Fix For: 0.17.0
>
> Attachments: flushToFsync.patch
>
>
> Starting fron hadoop-0.17, calling flush of dfs becomes very expensive.
> The current implementation of flush should be renamed as sync.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Commented: (HADOOP-3146) DFSOutputStream.flush should be
renamed as DFSOutputStream.fsync
Posted by "Hudson (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HADOOP-3146?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12585960#action_12585960 ]
Hudson commented on HADOOP-3146:
--------------------------------
Integrated in Hadoop-trunk #451 (See [http://hudson.zones.apache.org/hudson/job/Hadoop-trunk/451/])
> DFSOutputStream.flush should be renamed as DFSOutputStream.fsync
> ----------------------------------------------------------------
>
> Key: HADOOP-3146
> URL: https://issues.apache.org/jira/browse/HADOOP-3146
> Project: Hadoop Core
> Issue Type: Bug
> Components: dfs
> Reporter: Runping Qi
> Assignee: dhruba borthakur
> Priority: Blocker
> Fix For: 0.17.0
>
> Attachments: flushToFsync.patch, flushToFsync2.patch
>
>
> Starting fron hadoop-0.17, calling flush of dfs becomes very expensive.
> The current implementation of flush should be renamed as sync.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Commented: (HADOOP-3146) DFS flush should be light weighted
Posted by "dhruba borthakur (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HADOOP-3146?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12584022#action_12584022 ]
dhruba borthakur commented on HADOOP-3146:
------------------------------------------
I agree that even if the flush call in 0.17 is much more heavyweight that earlier, I still do not understand why it is causing the task to hang. I wonder if there is a bug in the implementation of DFSOutputStream.flush() that is causing the observed problems!
> DFS flush should be light weighted
> ----------------------------------
>
> Key: HADOOP-3146
> URL: https://issues.apache.org/jira/browse/HADOOP-3146
> Project: Hadoop Core
> Issue Type: Bug
> Components: dfs
> Reporter: Runping Qi
> Assignee: dhruba borthakur
> Priority: Blocker
> Fix For: 0.17.0
>
>
> Starting fron hadoop-0.17, calling flush of dfs becomes very expensive.
> The current implementation of flush should be renamed as sync.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Updated: (HADOOP-3146) DFSOutputStream.flush should be
renamed as DFSOutputStream.fsync
Posted by "dhruba borthakur (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HADOOP-3146?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
dhruba borthakur updated HADOOP-3146:
-------------------------------------
Attachment: flushToFsync2.patch
The previous patch did not change one unit test that was still invoking the old flush() API.
> DFSOutputStream.flush should be renamed as DFSOutputStream.fsync
> ----------------------------------------------------------------
>
> Key: HADOOP-3146
> URL: https://issues.apache.org/jira/browse/HADOOP-3146
> Project: Hadoop Core
> Issue Type: Bug
> Components: dfs
> Reporter: Runping Qi
> Assignee: dhruba borthakur
> Priority: Blocker
> Fix For: 0.17.0
>
> Attachments: flushToFsync.patch, flushToFsync2.patch
>
>
> Starting fron hadoop-0.17, calling flush of dfs becomes very expensive.
> The current implementation of flush should be renamed as sync.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Assigned: (HADOOP-3146) SequenceFile should not call flush
excessively
Posted by "dhruba borthakur (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HADOOP-3146?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
dhruba borthakur reassigned HADOOP-3146:
----------------------------------------
Assignee: dhruba borthakur
> SequenceFile should not call flush excessively
> ----------------------------------------------
>
> Key: HADOOP-3146
> URL: https://issues.apache.org/jira/browse/HADOOP-3146
> Project: Hadoop Core
> Issue Type: Bug
> Components: dfs
> Reporter: Runping Qi
> Assignee: dhruba borthakur
> Priority: Blocker
> Fix For: 0.17.0
>
>
> Starting fron hadoop-0.17, calling flush of dfs becomes very expensive.
> It does not make much sense to call flush inside append.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Updated: (HADOOP-3146) DFSOutputStream.flush should be
renamed as DFSOutputStream.fsync
Posted by "dhruba borthakur (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HADOOP-3146?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
dhruba borthakur updated HADOOP-3146:
-------------------------------------
Status: Patch Available (was: Open)
> DFSOutputStream.flush should be renamed as DFSOutputStream.fsync
> ----------------------------------------------------------------
>
> Key: HADOOP-3146
> URL: https://issues.apache.org/jira/browse/HADOOP-3146
> Project: Hadoop Core
> Issue Type: Bug
> Components: dfs
> Reporter: Runping Qi
> Assignee: dhruba borthakur
> Priority: Blocker
> Fix For: 0.17.0
>
> Attachments: flushToFsync.patch, flushToFsync2.patch
>
>
> Starting fron hadoop-0.17, calling flush of dfs becomes very expensive.
> The current implementation of flush should be renamed as sync.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Updated: (HADOOP-3146) DFSOutputStream.flush should be
renamed as DFSOutputStream.fsync
Posted by "dhruba borthakur (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HADOOP-3146?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
dhruba borthakur updated HADOOP-3146:
-------------------------------------
Hadoop Flags: [Reviewed]
Status: Patch Available (was: Open)
> DFSOutputStream.flush should be renamed as DFSOutputStream.fsync
> ----------------------------------------------------------------
>
> Key: HADOOP-3146
> URL: https://issues.apache.org/jira/browse/HADOOP-3146
> Project: Hadoop Core
> Issue Type: Bug
> Components: dfs
> Reporter: Runping Qi
> Assignee: dhruba borthakur
> Priority: Blocker
> Fix For: 0.17.0
>
> Attachments: flushToFsync.patch
>
>
> Starting fron hadoop-0.17, calling flush of dfs becomes very expensive.
> The current implementation of flush should be renamed as sync.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Updated: (HADOOP-3146) DFS flush should be light weighted
Posted by "Runping Qi (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HADOOP-3146?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Runping Qi updated HADOOP-3146:
-------------------------------
Description:
Starting fron hadoop-0.17, calling flush of dfs becomes very expensive.
The current implementation of flush should be renamed as sync.
was:
Starting fron hadoop-0.17, calling flush of dfs becomes very expensive.
It does not make much sense to call flush inside append.
Summary: DFS flush should be light weighted (was: SequenceFile should not call flush excessively)
> DFS flush should be light weighted
> ----------------------------------
>
> Key: HADOOP-3146
> URL: https://issues.apache.org/jira/browse/HADOOP-3146
> Project: Hadoop Core
> Issue Type: Bug
> Components: dfs
> Reporter: Runping Qi
> Assignee: dhruba borthakur
> Priority: Blocker
> Fix For: 0.17.0
>
>
> Starting fron hadoop-0.17, calling flush of dfs becomes very expensive.
> The current implementation of flush should be renamed as sync.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Updated: (HADOOP-3146) SequenceFile should not call flush
excessively
Posted by "Owen O'Malley (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HADOOP-3146?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Owen O'Malley updated HADOOP-3146:
----------------------------------
Component/s: (was: mapred)
dfs
Description:
Starting fron hadoop-0.17, calling flush of dfs becomes very expensive.
It does not make much sense to call flush inside append.
was:
Starting fron hadoop-0.17, calling flush of dfs becomes very expensive.
It does not make much sense to call flush inside append.
Actually, I would vote for HDFS to make flush cheaper again. Flush should *not* mean that your bytes have reached the disk. It should only mean that they have been sent to the operating system.
> SequenceFile should not call flush excessively
> ----------------------------------------------
>
> Key: HADOOP-3146
> URL: https://issues.apache.org/jira/browse/HADOOP-3146
> Project: Hadoop Core
> Issue Type: Bug
> Components: dfs
> Reporter: Runping Qi
>
> Starting fron hadoop-0.17, calling flush of dfs becomes very expensive.
> It does not make much sense to call flush inside append.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Updated: (HADOOP-3146) DFSOutputStream.flush should be
renamed as DFSOutputStream.fsync
Posted by "dhruba borthakur (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HADOOP-3146?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
dhruba borthakur updated HADOOP-3146:
-------------------------------------
Summary: DFSOutputStream.flush should be renamed as DFSOutputStream.fsync (was: DFS flush should be light weighted)
> DFSOutputStream.flush should be renamed as DFSOutputStream.fsync
> ----------------------------------------------------------------
>
> Key: HADOOP-3146
> URL: https://issues.apache.org/jira/browse/HADOOP-3146
> Project: Hadoop Core
> Issue Type: Bug
> Components: dfs
> Reporter: Runping Qi
> Assignee: dhruba borthakur
> Priority: Blocker
> Fix For: 0.17.0
>
>
> Starting fron hadoop-0.17, calling flush of dfs becomes very expensive.
> The current implementation of flush should be renamed as sync.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Commented: (HADOOP-3146) DFSOutputStream.flush should be
renamed as DFSOutputStream.fsync
Posted by "Raghu Angadi (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HADOOP-3146?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12584945#action_12584945 ]
Raghu Angadi commented on HADOOP-3146:
--------------------------------------
+1.
> DFSOutputStream.flush should be renamed as DFSOutputStream.fsync
> ----------------------------------------------------------------
>
> Key: HADOOP-3146
> URL: https://issues.apache.org/jira/browse/HADOOP-3146
> Project: Hadoop Core
> Issue Type: Bug
> Components: dfs
> Reporter: Runping Qi
> Assignee: dhruba borthakur
> Priority: Blocker
> Fix For: 0.17.0
>
> Attachments: flushToFsync.patch
>
>
> Starting fron hadoop-0.17, calling flush of dfs becomes very expensive.
> The current implementation of flush should be renamed as sync.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Updated: (HADOOP-3146) DFSOutputStream.flush should be
renamed as DFSOutputStream.fsync
Posted by "dhruba borthakur (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HADOOP-3146?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
dhruba borthakur updated HADOOP-3146:
-------------------------------------
Resolution: Fixed
Status: Resolved (was: Patch Available)
I just committed this.
> DFSOutputStream.flush should be renamed as DFSOutputStream.fsync
> ----------------------------------------------------------------
>
> Key: HADOOP-3146
> URL: https://issues.apache.org/jira/browse/HADOOP-3146
> Project: Hadoop Core
> Issue Type: Bug
> Components: dfs
> Reporter: Runping Qi
> Assignee: dhruba borthakur
> Priority: Blocker
> Fix For: 0.17.0
>
> Attachments: flushToFsync.patch, flushToFsync2.patch
>
>
> Starting fron hadoop-0.17, calling flush of dfs becomes very expensive.
> The current implementation of flush should be renamed as sync.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Commented: (HADOOP-3146) DFSOutputStream.flush should be
renamed as DFSOutputStream.fsync
Posted by "Hadoop QA (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HADOOP-3146?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12585454#action_12585454 ]
Hadoop QA commented on HADOOP-3146:
-----------------------------------
+1 overall. Here are the results of testing the latest attachment
http://issues.apache.org/jira/secure/attachment/12379324/flushToFsync2.patch
against trunk revision 643282.
@author +1. The patch does not contain any @author tags.
tests included +1. The patch appears to include 9 new or modified tests.
javadoc +1. The javadoc tool did not generate any warning messages.
javac +1. The applied patch does not generate any new javac compiler warnings.
release audit +1. The applied patch does not generate any new release audit warnings.
findbugs +1. The patch does not introduce any new Findbugs warnings.
core tests +1. The patch passed core unit tests.
contrib tests +1. The patch passed contrib unit tests.
Test results: http://hudson.zones.apache.org/hudson/job/Hadoop-Patch/2155/testReport/
Findbugs warnings: http://hudson.zones.apache.org/hudson/job/Hadoop-Patch/2155/artifact/trunk/build/test/findbugs/newPatchFindbugsWarnings.html
Checkstyle results: http://hudson.zones.apache.org/hudson/job/Hadoop-Patch/2155/artifact/trunk/build/test/checkstyle-errors.html
Console output: http://hudson.zones.apache.org/hudson/job/Hadoop-Patch/2155/console
This message is automatically generated.
> DFSOutputStream.flush should be renamed as DFSOutputStream.fsync
> ----------------------------------------------------------------
>
> Key: HADOOP-3146
> URL: https://issues.apache.org/jira/browse/HADOOP-3146
> Project: Hadoop Core
> Issue Type: Bug
> Components: dfs
> Reporter: Runping Qi
> Assignee: dhruba borthakur
> Priority: Blocker
> Fix For: 0.17.0
>
> Attachments: flushToFsync.patch, flushToFsync2.patch
>
>
> Starting fron hadoop-0.17, calling flush of dfs becomes very expensive.
> The current implementation of flush should be renamed as sync.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Updated: (HADOOP-3146) SequenceFile should not call flush
excessively
Posted by "dhruba borthakur (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HADOOP-3146?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
dhruba borthakur updated HADOOP-3146:
-------------------------------------
Fix Version/s: 0.17.0
Priority: Blocker (was: Major)
> SequenceFile should not call flush excessively
> ----------------------------------------------
>
> Key: HADOOP-3146
> URL: https://issues.apache.org/jira/browse/HADOOP-3146
> Project: Hadoop Core
> Issue Type: Bug
> Components: dfs
> Reporter: Runping Qi
> Priority: Blocker
> Fix For: 0.17.0
>
>
> Starting fron hadoop-0.17, calling flush of dfs becomes very expensive.
> It does not make much sense to call flush inside append.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Commented: (HADOOP-3146) DFS flush should be light weighted
Posted by "Raghu Angadi (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HADOOP-3146?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12584014#action_12584014 ]
Raghu Angadi commented on HADOOP-3146:
--------------------------------------
How expensive really is current flush()?
Of course, lighter flush() is better. I think we should do this irrespective of how much it saves.
> DFS flush should be light weighted
> ----------------------------------
>
> Key: HADOOP-3146
> URL: https://issues.apache.org/jira/browse/HADOOP-3146
> Project: Hadoop Core
> Issue Type: Bug
> Components: dfs
> Reporter: Runping Qi
> Assignee: dhruba borthakur
> Priority: Blocker
> Fix For: 0.17.0
>
>
> Starting fron hadoop-0.17, calling flush of dfs becomes very expensive.
> The current implementation of flush should be renamed as sync.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Updated: (HADOOP-3146) DFSOutputStream.flush should be
renamed as DFSOutputStream.fsync
Posted by "dhruba borthakur (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HADOOP-3146?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
dhruba borthakur updated HADOOP-3146:
-------------------------------------
Attachment: flushToFsync.patch
Renamed DFSOutputStream.flush() to DFSOutputStream.fsync(). This ensures that programs that were invoking OutputStream.flush() lavishly would continue to work as before.
> DFSOutputStream.flush should be renamed as DFSOutputStream.fsync
> ----------------------------------------------------------------
>
> Key: HADOOP-3146
> URL: https://issues.apache.org/jira/browse/HADOOP-3146
> Project: Hadoop Core
> Issue Type: Bug
> Components: dfs
> Reporter: Runping Qi
> Assignee: dhruba borthakur
> Priority: Blocker
> Fix For: 0.17.0
>
> Attachments: flushToFsync.patch
>
>
> Starting fron hadoop-0.17, calling flush of dfs becomes very expensive.
> The current implementation of flush should be renamed as sync.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.