[
https://issues.apache.org/jira/browse/BOOKKEEPER-744?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13984635#comment-13984635
]
Rakesh R commented on BOOKKEEPER-744:
-------------------------------------
Thanks [~ikelly] for the patch. I've few minor comments:
# Please add unit in the description.
The interval is set in seconds.
+#auditorPeriodicBookieCheckInterval=84600
# We could skip the current bookie checking cycle if the replication is disabled
{code}
if (!ledgerUnderreplicationManager.isLedgerReplicationEnabled()) {
LOG.info("Ledger replication disabled, skipping");
return;
}
{code}
# Improve logs:
a)
Now we have two periodic checks 'AuditorPeriodicCheckInterval and
AuditorPeriodicBookieCheckInterval. It would be good to print the interval in
the log messages to show the periods.
{code}
LOG.info("Auditor periodic ledger checking enabled
'auditorPeriodicCheckInterval' {} seconds", interval);
LOG.info("Auditor periodic bookie checking enabled
'auditorPeriodicBookieCheckInterval' {} seconds", interval);
{code}
b) Add disable log message too
{code}
LOG.info("Periodic bookie checking disabled");
{code}
# Test case took 4secs in my env. IMHO, good to increase timeout=5000 to
slightly bigger value say 30000. In worst case if the h/w is slow we can handle
that case and still give the test a chance to pass.
-Rakesh
> Run the auditor bookie check periodically
> -----------------------------------------
>
> Key: BOOKKEEPER-744
> URL: https://issues.apache.org/jira/browse/BOOKKEEPER-744
> Project: Bookkeeper
> Issue Type: Bug
> Components: bookkeeper-auto-recovery
> Reporter: Ivan Kelly
> Assignee: Ivan Kelly
> Fix For: 4.3.0, 4.2.3
>
> Attachments: 0001-Run-the-bookie-check-periodically.trunk.patch,
> 0003-Run-the-bookie-check-periodically.trunk.patch,
> 0005-Run-the-bookie-check-periodically.branch4.2.patch
>
>
> Previous the bookie check only runs when a bookie fails. If for some reason
> this doesn't pick up a failure, the failure will go undetected until the next
> time a bookie fails, in which case quorum could have been lost. This fix
> makes the bookie check run periodically, by default once a day.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
