cassandra-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Marcus Eriksson (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (CASSANDRA-11172) Infinite loop bug adding high-level SSTableReader in compaction
Date Wed, 17 Feb 2016 08:45:18 GMT

    [ https://issues.apache.org/jira/browse/CASSANDRA-11172?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15150140#comment-15150140
] 

Marcus Eriksson commented on CASSANDRA-11172:
---------------------------------------------

The problem is (probably) that we have an sstable in the wrong level in the manifest.

I have not been able to reproduce this locally but it would make sense (ie, the sstable would
not get removed from the manifest and we could end up in an infinite loop), but I think we
should see an exception before things explode like this, so it would be really helpful if
anyone has logs an hour or so *before* this happens

> Infinite loop bug adding high-level SSTableReader in compaction
> ---------------------------------------------------------------
>
>                 Key: CASSANDRA-11172
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-11172
>             Project: Cassandra
>          Issue Type: Bug
>          Components: Compaction
>         Environment: DSE 4.x / Cassandra 2.1.11.969
>            Reporter: Jeff Ferland
>            Assignee: Marcus Eriksson
>         Attachments: beep.txt, tpstats_compaction.txt, trapped_in_compaction.txt, trapped_in_compaction_mixed.txt
>
>
> Observed that after a large repair on LCS that sometimes the system will enter an infinite
loop with vast amounts of logs lines recording, "Adding high-level (L${LEVEL}) SSTableReader(path='${TABLE}')
to candidates"
> This results in an outage of the node and eventual crashing. The log spam quickly rotates
out possibly useful earlier debugging.



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

Mime
View raw message