[
https://issues.apache.org/jira/browse/HDFS-7754?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14312994#comment-14312994
]
Haohui Mai commented on HDFS-7754:
----------------------------------
bq. IS2_INCONSISTENT_SYNC's has false alarm issue as described in its doc.
I'm unsure the value of introducing a (possible benign) race either. Does the
race improve the performance significantly? Does it sit on the critical path?
I think that the values of the time of developer / maintainers are
under-appreciated. Note that it takes the time of developers / maintainers to
reason about it even if it's a false alarm -- unfortunately it is error-prone
and ineffective [1]. Arguably putting the related code in a critical section is
not the greatest fix, but given that fact that it is not in a
performance-critical path I think it is okay to fix the race and to save the
time of every developers and maintainers. If it is indeed a performance issue
we can revisit it.
Reference:
1. S. Savage et al.. Eraser: A Dynamic Data Race Detector for Multithreaded
Programs. In SOSP'97.
> Fix findbugs warning produced by HDFS-7710
> ------------------------------------------
>
> Key: HDFS-7754
> URL: https://issues.apache.org/jira/browse/HDFS-7754
> Project: Hadoop HDFS
> Issue Type: Bug
> Affects Versions: 2.7.0
> Reporter: Akira AJISAKA
> Assignee: Akira AJISAKA
> Attachments: HDFS-7710-001.patch
>
>
> There is a findbugs warning produced by HDFS-7710.
> https://builds.apache.org/job/PreCommit-HDFS-Build/9493//artifact/patchprocess/newPatchFindbugsWarningshadoop-hdfs.html
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
