accumulo-notifications mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Eric Newton (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (ACCUMULO-3618) Tests which invoke `accumulo admin stop ...` fail
Date Tue, 24 Feb 2015 19:11:05 GMT

    [ https://issues.apache.org/jira/browse/ACCUMULO-3618?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14335265#comment-14335265
] 

Eric Newton commented on ACCUMULO-3618:
---------------------------------------

The TabletStateChangeIterator wasn't returning tablets because it the tablet was online for
an online table.  However, the master goal state needs to be taken into account, too.

So, the TabletStateChangeIterator needs to know more information:

* servers shutting down
* migrations
* master state change

It's not that we can't fix it (I've done it for 1.7), but it *feels* like a pretty big change
for 1.5 and 1.6.

> Tests which invoke `accumulo admin stop ...` fail
> -------------------------------------------------
>
>                 Key: ACCUMULO-3618
>                 URL: https://issues.apache.org/jira/browse/ACCUMULO-3618
>             Project: Accumulo
>          Issue Type: Bug
>          Components: test
>            Reporter: Josh Elser
>            Assignee: Eric Newton
>            Priority: Critical
>             Fix For: 1.7.0
>
>
> I'm noticing that all of the tests which are stopping a tabletserver gracefully are failing
for me:
> * SslWithClientAuthIT.adminStop
> * ShutdownIT.adminStop
> * SslIT.adminStop
> It failed locally in addition to on a Jenkins instance. I dug into it locally a little
bit: the FATE operation (ShutdownTServer) never finishes because the TabletServer which is
requested to be shutdown never gets all of its tablets unassigned.
> I haven't been able to figure out why exactly the master didn't move all of the tablets
away.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message