[
https://issues.apache.org/jira/browse/CASSANDRA-5351?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13795684#comment-13795684
]
Jonathan Ellis commented on CASSANDRA-5351:
-------------------------------------------
bq. I'm not sure if bootstrapping data can be considered anti-compacted through
cummutativity
Yes. Since we only stream data from a single token range, we don't need to
worry about mixing state for different peers. If the data for sstable X was
repaired on replica A vs replicas B and C, the data will still be repaired when
moved to A'. (We would have to teach streaming to not mix repaired/unrepaired
sstables so that state can be preserved on the target.)
bq. C is forced to do an -ipr repair as A and B have already anti-compacted
I think it will be simpler if we allow C to do a non-ipr repair (and overstream
previously repaired data, since A and B won't include it in their MT). This is
a corner case that will only happen if C crashes after streaming has completed
successfully but before the repair is marked successful.
bq. Local DC repair vs. global is largely incompatible with this.
Yes, I think local implicitly means -ipr. (So, if you are fond of local
repair, you are no worse off.)
> Avoid repairing already-repaired data by default
> ------------------------------------------------
>
> Key: CASSANDRA-5351
> URL: https://issues.apache.org/jira/browse/CASSANDRA-5351
> Project: Cassandra
> Issue Type: Task
> Components: Core
> Reporter: Jonathan Ellis
> Assignee: Lyuben Todorov
> Labels: repair
> Fix For: 2.1
>
>
> Repair has always built its merkle tree from all the data in a columnfamily,
> which is guaranteed to work but is inefficient.
> We can improve this by remembering which sstables have already been
> successfully repaired, and only repairing sstables new since the last repair.
> (This automatically makes CASSANDRA-3362 much less of a problem too.)
> The tricky part is, compaction will (if not taught otherwise) mix repaired
> data together with non-repaired. So we should segregate unrepaired sstables
> from the repaired ones.
--
This message was sent by Atlassian JIRA
(v6.1#6144)
