[
https://issues.apache.org/jira/browse/ACCUMULO-681?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13410370#comment-13410370
]
Eric Newton commented on ACCUMULO-681:
--------------------------------------
Strange... here's the master logs, where the tablet is assigned to node4
{noformat}
10 12:00:02,535 [master.Master] DEBUG: Root Tablet assigning tablet
!0;!0<<=192.168.117.11:9997[2383418e9a17f4b]
10 12:00:09,463 [master.Master] DEBUG: Root Tablet assigning tablet
!0;!0<<=192.168.117.11:9997[2383418e9a17f4b]
10 12:00:09,548 [master.Master] DEBUG: Root Tablet assigning tablet
!0;!0<<=192.168.117.5:9997[2383418e9a17f44]
{noformat}
So, if node10 processed a late message, and saw a cached or unsynched
future_location set to itself, it might open the root tablet at the same time
as node4.
> close consistency check failure
> -------------------------------
>
> Key: ACCUMULO-681
> URL: https://issues.apache.org/jira/browse/ACCUMULO-681
> Project: Accumulo
> Issue Type: Bug
> Affects Versions: 1.5.0-SNAPSHOT
> Environment: 10 node test cluster running randomwalk Concurrent graph
> Reporter: Eric Newton
> Assignee: Eric Newton
> Priority: Blocker
> Fix For: 1.5.0
>
>
> I recently added tablet server admin shutdown to the Concurrent randomwalk
> test.
> After an hour, the following table problem appeared:
> {noformat}
> 10 12:00:21,755 [tabletserver.Tablet] ERROR: Data file in !METADATA differ
> from in memory data !0;!0<<
> {noformat}
> Note that this is the root tablet, so some other process was updating the
> files behind this server's back; I suspect inconsistent assignment.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
