[
https://issues.apache.org/jira/browse/HBASE-20069?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16375980#comment-16375980
]
stack commented on HBASE-20069:
-------------------------------
.002 addresses [~mdrob] review and more sketchyness on my part was found by
[~chia7712] up on rb.
[~mdrob]
bq. I think the dead store in MutliButeBuf needs to be addressed by using it on
line 297. Probably need to review the logic there more closely to potentially
match lines 233-5.
You are right. Added fix and test. Thanks. Messing around, seems like more fix
and test needed in here beyond getShort. Filed HBASE-20078.
bq. In CleanerChore, I think we want a lazy init singleton pattern.
What is going on in CleanerChore is wonky. Singleton would be nice but then
later, we want to be able to change the pool down in onConfigurationChange
event.
bq. WakeupFlushToken could be a final class, I think. We're trying to emulate a
singleton without using an enum? should add a private constructor and have a
static instance variable. equals can check for reference equality then.
Tried your suggestion. We save on a few class constructions which is good.
Findbugs looking for hashCode now too... Its not pretty.
bq. I don't like us manually throwing NPE in AsyncFSWAL. Better would be
something like writer = Preconditions.checkNotNull(nextWriter);
Did as you suggested. Thanks.
[~chia7712]
Thanks. I took a look and just undid the synchronization altogether.
> fix existing findbugs errors in hbase-server
> --------------------------------------------
>
> Key: HBASE-20069
> URL: https://issues.apache.org/jira/browse/HBASE-20069
> Project: HBase
> Issue Type: Sub-task
> Components: findbugs
> Reporter: Sean Busbey
> Assignee: stack
> Priority: Critical
> Attachments: FindBugs Report.htm, HBASE-20069.branch-2.001.patch,
> HBASE-20069.branch-2.002.patch
>
>
> now that findbugs is running on precommit we have some cleanup to do.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
