[
https://issues.apache.org/jira/browse/CASSANDRA-7357?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14627291#comment-14627291
]
Paulo Motta commented on CASSANDRA-7357:
----------------------------------------
hmm, nevermind, after closer inspection I verified repair tracing is an
optional feature so we can't rely on that for cleaning up repair leftover
snapshots.
the simplest solution here would be to add a "repair-" prefix to snapshots
originated from repair, but then this could delete user snapshots with this
same prefix. other options include adding a flag to the snapshot indicating it
originated from a repair session, or persist somewhere repairs in progress and
clean up snapshots from these repairs on startup (this would be the most
elegant solution I guess).
> Cleaning up snapshots on startup leftover from repairs that got interrupted
> ---------------------------------------------------------------------------
>
> Key: CASSANDRA-7357
> URL: https://issues.apache.org/jira/browse/CASSANDRA-7357
> Project: Cassandra
> Issue Type: Bug
> Components: Core
> Reporter: Stephen Johnson
> Assignee: Paulo Motta
> Priority: Minor
> Fix For: 2.1.x
>
>
> When a repair is interrupted, such as a restart, it can leave a snapshot
> behind on disk. These could be checked for and cleared possibly on startup.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
