[
https://issues.apache.org/jira/browse/HBASE-6388?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13468532#comment-13468532
]
ramkrishna.s.vasudevan commented on HBASE-6388:
-----------------------------------------------
{code}
if (!this.killed && this.fsOk) {
waitOnAllRegionsToClose(abortRequested);
LOG.info("stopping server " + this.serverNameFromMasterPOV +
"; all regions closed.");
}
//fsOk flag may be changed when closing regions throws exception.
if (!this.killed && this.fsOk) {
closeWAL(abortRequested ? false : true);
}
{code}
I think WAL closing is fine but the closing is not done parallel here. Do we
need to address parallelizing closes alone then? What you feel Stack?
> Avoid potential data loss if the flush fails during regionserver shutdown
> -------------------------------------------------------------------------
>
> Key: HBASE-6388
> URL: https://issues.apache.org/jira/browse/HBASE-6388
> Project: HBase
> Issue Type: Bug
> Affects Versions: 0.89-fb
> Reporter: Amitanand Aiyer
> Assignee: Amitanand Aiyer
> Priority: Critical
> Fix For: 0.96.0
>
> Attachments:
> 0001-HBASE-6388-89-fb-parallelize-close-and-avoid-deletin.patch
>
>
> During a controlled shutdown, Regionserver deletes HLogs even if
> HRegion.close() fails. We should not be doing this.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
