cassandra-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Peter Schuller (JIRA)" <>
Subject [jira] [Commented] (CASSANDRA-3882) avoid distributed deadlock in migration stage
Date Wed, 23 May 2012 02:50:41 GMT


Peter Schuller commented on CASSANDRA-3882:

Yes, that is the issue I'm seeing. And I completely agree about synchronous (with respect
to other nodes) operations on stages.

I also agree that I can't think of why specifically this would be unsafe to do asynchronously;
I was just taking the ultra-conservative approach.

Looked through the patch very briefly and it seems reasonable, but I haven't looked at it
carefully or tested it. One thing I'm concerned with for large clusters is the total number
of messages back and fourth, since I believe it ends up being quadratic in cluster size. That
said, maybe that's okay up to at least several hundred nodes.
> avoid distributed deadlock in migration stage
> ---------------------------------------------
>                 Key: CASSANDRA-3882
>                 URL:
>             Project: Cassandra
>          Issue Type: Bug
>          Components: Core
>    Affects Versions: 1.1.0
>            Reporter: Peter Schuller
>            Assignee: Richard Low
>             Fix For: 1.1.1
>         Attachments: CASSANDRA-3882-async.patch, CASSANDRA-3882-hack.txt
> This is follow-up work for the remainders of CASSANDRA-3832 which was only a partial
fix. The deadlock in the migration stage needs to be fixed, as it can cause bootstrap (at
least) to take potentially a very very long time to complete, and might also cause a lack
of schema propagation until otherwise "poked".

This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:!default.jspa
For more information on JIRA, see:


View raw message