cassandra-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Benedict (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (CASSANDRA-7066) Simplify (and unify) cleanup of compaction leftovers
Date Mon, 17 Aug 2015 09:25:46 GMT

    [ https://issues.apache.org/jira/browse/CASSANDRA-7066?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14699251#comment-14699251
] 

Benedict commented on CASSANDRA-7066:
-------------------------------------

Thanks. I've committed as 5726625a5c7cf47a67509540f7146d05b668bc20.

There is one more thing to consider, though, which I did not want to hold up commit for: in
the event of file system failure, and our pessimistically assuming we should not rollback
the transaction, we can (and probably will) leave _incomplete_ sstables. This isn't a huge
hurdle given the encountering of a major file system failure, and the user can easily delete
this file, however I wonder if we should not try to read the end of the file on startup, and
if we cannot to stash it somewhere for the operator to use if they want, but to prevent it
being loaded on startup in future. This should perhaps be a follow up ticket, but I feel it
at least warrants a brief discussion here.

> Simplify (and unify) cleanup of compaction leftovers
> ----------------------------------------------------
>
>                 Key: CASSANDRA-7066
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-7066
>             Project: Cassandra
>          Issue Type: Improvement
>          Components: Core
>            Reporter: Benedict
>            Assignee: Stefania
>            Priority: Minor
>              Labels: benedict-to-commit, compaction
>             Fix For: 3.0 alpha 1
>
>         Attachments: 7066.txt
>
>
> Currently we manage a list of in-progress compactions in a system table, which we use
to cleanup incomplete compactions when we're done. The problem with this is that 1) it's a
bit clunky (and leaves us in positions where we can unnecessarily cleanup completed files,
or conversely not cleanup files that have been superceded); and 2) it's only used for a regular
compaction - no other compaction types are guarded in the same way, so can result in duplication
if we fail before deleting the replacements.
> I'd like to see each sstable store in its metadata its direct ancestors, and on startup
we simply delete any sstables that occur in the union of all ancestor sets. This way as soon
as we finish writing we're capable of cleaning up any leftovers, so we never get duplication.
It's also much easier to reason about.



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

Mime
View raw message