accumulo-notifications mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "ASF subversion and git services (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (ACCUMULO-2621) Masters not restarting during concurrent randomwalk
Date Tue, 15 Apr 2014 18:44:15 GMT

    [ https://issues.apache.org/jira/browse/ACCUMULO-2621?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13969877#comment-13969877
] 

ASF subversion and git services commented on ACCUMULO-2621:
-----------------------------------------------------------

Commit ef12e59ecf918299574eef94deb62464fb3f54bd in accumulo's branch refs/heads/1.6.0-SNAPSHOT
from [~bhavanki]
[ https://git-wip-us.apache.org/repos/asf?p=accumulo.git;h=ef12e59 ]

ACCUMULO-2621 Add delay to Shutdown step of continuous randomwalk, update README

The Concurrent.xml randomwalk module can shut down and restart the Accumulo cluster during
the test. This commit adds a 10 second delay after shutdown, to give the master plenty of
time to exit before the test attempts to start it up again.

Also, it is essential that the walker(s) running the test have passwordless SSH access to
the cluster to run $ACCUMULO_HOME/bin/start-all.sh. Otherwise, the masters will not restart
and the test will eventually get stuck. The randomwalk README.md now has information about
that.


> Masters not restarting during concurrent randomwalk
> ---------------------------------------------------
>
>                 Key: ACCUMULO-2621
>                 URL: https://issues.apache.org/jira/browse/ACCUMULO-2621
>             Project: Accumulo
>          Issue Type: Test
>          Components: test
>            Reporter: Bill Havanki
>            Assignee: Bill Havanki
>            Priority: Critical
>              Labels: 16_qa_bug, randomwalk, test
>             Fix For: 1.6.1
>
>         Attachments: ACCUMULO-2621.v1.patch.txt
>
>
> The Concurrent randomwalk test can stop and restart the masters. Under 1.6.0-SNAPSHOT,
the stopped masters are not restarting, and eventually the test becomes stuck reporting "No
matchers..." forever.
> Tested on 7-node CentOS 6.4 cluster, 2 masters. The active master seems to die first,
then the standby that becomes the new master.



--
This message was sent by Atlassian JIRA
(v6.2#6252)

Mime
View raw message