cassandra-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Paulo Motta (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (CASSANDRA-14079) Prevent compaction strategies from looping indefinitely
Date Thu, 30 Nov 2017 12:09:00 GMT

    [ https://issues.apache.org/jira/browse/CASSANDRA-14079?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16272581#comment-16272581
] 

Paulo Motta commented on CASSANDRA-14079:
-----------------------------------------

bq. +1, but it was not because of CASSANDRA-13948 right? 

After the disk boundaries changed, SSTables were not being removed from the correct strategy
after they were compacted because their indexes changed, so they were re-considered for compaction
but the strategy was not able to acquire the lock from the tracker, and so {{getNextBackgroundTask}}
was looping indefinitely trying to choose the same sstables as candidates which were already
compacted.

> Prevent compaction strategies from looping indefinitely
> -------------------------------------------------------
>
>                 Key: CASSANDRA-14079
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-14079
>             Project: Cassandra
>          Issue Type: Improvement
>            Reporter: Paulo Motta
>            Assignee: Paulo Motta
>            Priority: Minor
>
> As a result of CASSANDRA-13948, LCS was looping indefinitely trying to generate the same
candidates for SSTables which were not on the tracker.
> We should add a protection on compaction strategies against looping indefinitely to avoid
similar bugs in the future.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)

---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscribe@cassandra.apache.org
For additional commands, e-mail: commits-help@cassandra.apache.org


Mime
View raw message