[
https://issues.apache.org/jira/browse/CASSANDRA-1546?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12917567#action_12917567
]
Kelvin Kakugawa commented on CASSANDRA-1546:
--------------------------------------------
I'm a little wary of having "bad" state on disk, because it introduces
additional variables that have to be reasoned about.
As of 2 months ago, AES doesn't perform a read-only major compaction like it
did in 0.6. It streams across the relevant subsections of each SST on disk to
the target node, which is concatenated into a new SST. So, if:
1) A streams an AES repair SST to B,
2) that streamed SST is never transformed (e.g. via compaction), then
3) B streams an AES repair SST back to A.
It will send the untransformed columns from A, back to A. When I have time,
I'll re-read the compaction code to make sure this hasn't changed.
I believe you're right that #1072 also suffers from this potential race
condition during reads, but I'd have to review that code path.
I'm definitely in favor of moving the metadata into a different CF. Of
interest, since all the columns have the same TTL, a modified compaction
strategy could be implemented that completely deletes the SSTs for this CF w/o
inspection based on when the SST was written.
I'll read through your v2 patch. Thanks for rebasing.
> (Yet another) approach to counting
> ----------------------------------
>
> Key: CASSANDRA-1546
> URL: https://issues.apache.org/jira/browse/CASSANDRA-1546
> Project: Cassandra
> Issue Type: New Feature
> Components: Core
> Reporter: Sylvain Lebresne
> Assignee: Sylvain Lebresne
> Fix For: 0.7.0
>
> Attachments: 0001-Remove-IClock-from-internals.patch,
> 0001-v2-Remove-IClock-from-internals.patch, 0002-Counters.patch,
> 0002-v2-Counters.patch, 0003-Generated-thrift-files-changes.patch,
> 0003-v2-Thrift-changes.patch
>
>
> This could be described as a mix between CASSANDRA-1072 without clocks and
> CASSANDRA-1421.
> More details in the comment below.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
