[
https://issues.apache.org/jira/browse/HBASE-3343?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12973471#action_12973471
]
stack commented on HBASE-3343:
------------------------------
@jdcryans +1 on your patch though its addressing other than whats in this log;
log is about an aborted server not going down. Code looks like this in the
abort processing:
{code}
....
629 if (this.killed) {
630 // Just skip out w/o closing regions.
631 } else if (abortRequested) {
632 if (this.fsOk) {
633 closeAllRegions(abortRequested); // Don't leave any open file
handles
634 closeWAL(false);
635 }
636 LOG.info("aborting server at: " + this.serverInfo.getServerName());
637 } else {
638 closeAllRegions(abortRequested);
639 closeWAL(true);
640 closeAllScanners();
641 LOG.info("stopping server at: " + this.serverInfo.getServerName());
642 }
643 // Interrupt catalog tracker here in case any regions being opened out
in
644 // handlers are stuck waiting on meta or root.
645 if (this.catalogTracker != null) this.catalogTracker.stop();
646 waitOnAllRegionsToClose();
....
{code}
... so if an abort is requested AND the fs is NOT OK, then we won't close
regions... we just skip out. ONLY, we then fall into waitOnAllRegionsToClose
on line #646 above which will never complete because we didn't do close on
regions.
I think simplest fix is adding this:
{code}
646 if (this.fsOK) waitOnAllRegionsToClose();
{code}
Let me commit your patch and mine together.
> Server not shutting down after losing log lease
> -----------------------------------------------
>
> Key: HBASE-3343
> URL: https://issues.apache.org/jira/browse/HBASE-3343
> Project: HBase
> Issue Type: Bug
> Affects Versions: 0.90.0
> Reporter: Todd Lipcon
> Assignee: stack
> Priority: Critical
> Fix For: 0.90.0
>
> Attachments: HBASE-3343.patch, shutdown-logs.txt.bz2, stuck-server.txt
>
>
> Ran into this bug testing 0.90rc2. I kill -STOPed a server, and then -CONT it
> after its logs had been split. It correctly decided it should abort, but got
> stuck during the shutdown process.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
