[
https://issues.apache.org/jira/browse/CASSANDRA-5220?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13971595#comment-13971595
]
Yuki Morishita commented on CASSANDRA-5220:
-------------------------------------------
Thanks, Ryan.
Time increase in Incoming/OutboundTcpConnection indicate repair is spending
more time in messaging.
It is understandable the messaging is taking more than 200x for repairing 256x
ranges.
One possible solutin is to repair multiple ranges at once.
I have two ideas in my mind:
# Build two-level MerkleTree of multiple ranges. In the lower level we have
regular, per range MT and in the upper level, we have MT whose leaf is root
hash of lower MT. So we can carry multiple MT in one round trip of message.
# Send validation request once for all ranges, replica node builds MT for each
range one by one, and sent back MT as it is built.
> Repair improvements when using vnodes
> -------------------------------------
>
> Key: CASSANDRA-5220
> URL: https://issues.apache.org/jira/browse/CASSANDRA-5220
> Project: Cassandra
> Issue Type: Improvement
> Components: Core
> Affects Versions: 1.2.0 beta 1
> Reporter: Brandon Williams
> Assignee: Yuki Morishita
> Labels: performance, repair
> Fix For: 2.1 beta2
>
> Attachments: 5220-yourkit.png, 5220-yourkit.tar.bz2
>
>
> Currently when using vnodes, repair takes much longer to complete than
> without them. This appears at least in part because it's using a session per
> range and processing them sequentially. This generates a lot of log spam
> with vnodes, and while being gentler and lighter on hard disk deployments,
> ssd-based deployments would often prefer that repair be as fast as possible.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
