cassandra-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Stu Hood (JIRA)" <>
Subject [jira] Commented: (CASSANDRA-193) Proactive repair
Date Wed, 24 Jun 2009 17:15:07 GMT


Stu Hood commented on CASSANDRA-193:

> ...what is the behavior after a fresh start if a Repair phase begins?
Well, you would have a single large invalid range (or some number of smaller ranges summing
to the full range), which could be fetched sequentially from disk.

Rather than being completely lazy, (depending on how much extra load it would cause) we could
hook the AEService into compactions that are happening for other reasons, so that before the
compaction begins, the compactor fetches the current list of invalid ranges and fills them
in based on the merged data. I'm not sure how much of a win this would be, since we probably
don't want to slow down compactions, but if they aren't CPU bound, then it shouldn't hurt.

> Proactive repair
> ----------------
>                 Key: CASSANDRA-193
>                 URL:
>             Project: Cassandra
>          Issue Type: New Feature
>            Reporter: Jonathan Ellis
>            Assignee: Stu Hood
>             Fix For: 0.5
> Currently cassandra supports "read repair," i.e., lazy repair when a read is done.  This
is better than nothing but is not sufficient for some cases (e.g. catastrophic node failure
where you need to rebuild all of a node's data on a new machine).
> Dynamo uses merkle trees here.  This is harder for Cassandra given the CF data model
but I suppose we could just hash the serialized CF value.

This message is automatically generated by JIRA.
You can reply to this email to add a comment to the issue online.

View raw message