[
https://issues.apache.org/jira/browse/CASSANDRA-2816?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13054297#comment-13054297
]
Sylvain Lebresne commented on CASSANDRA-2816:
---------------------------------------------
I'm not sure what you mean by "snapshotting immediately" or "polishing our
snapshot support", but one approach that I think is equivalent to that (or
maybe that is what you meant by 'snapshotting') would be to grab references to
the sstables at the very beginning for each request and use those all
throughout the repair. This has however a problem: this means we retain
sstables from being deleted during repair, including sstables that are
compacted in the meantime. Because repair can take a while, this will be bad.
This will also require changes to the wire protocol (because we'll need a way
to indicate during streaming the set of sstables to consider), and since we've
kind of decided to not do that in minor releases (at least until we've
discussed that), this means this cannot be released quickly. Which is bad,
because I'm pretty sure this is a good part of the reason why some people with
big data sets have had huge pain with repair.
Scheduling the validation one by one avoids those problems. In theory this
means we'll do less work in parallel, but in practice I doubt this is a big
since the goal is probably to have repair have less impact on the node rather
than more. It will also make this more easy to reason about.
> Repair doesn't synchronize merkle tree creation properly
> --------------------------------------------------------
>
> Key: CASSANDRA-2816
> URL: https://issues.apache.org/jira/browse/CASSANDRA-2816
> Project: Cassandra
> Issue Type: Bug
> Components: Core
> Affects Versions: 0.7.0, 0.8.0
> Reporter: Sylvain Lebresne
> Assignee: Sylvain Lebresne
> Labels: repair
>
> Being a little slow, I just realized after having opened CASSANDRA-2811 and
> CASSANDRA-2815 that there is a more general problem with repair.
> When a repair is started, it will send a number of merkle tree to its
> neighbor as well as himself and assume for correction that the building of
> those trees will be started on every node roughly at the same time (if not,
> we end up comparing data snapshot at different time and will thus mistakenly
> repair a lot of useless data). This is bogus for many reasons:
> * Because validation compaction runs on the same executor that other
> compaction, the start of the validation on the different node is subject to
> other compactions. 0.8 mitigates this in a way by being multi-threaded (and
> thus there is less change to be blocked a long time by a long running
> compaction), but the compaction executor being bounded, its still a problem)
> * if you run a nodetool repair without arguments, it will repair every CFs.
> As a consequence it will generate lots of merkle tree requests and all of
> those requests will be issued at the same time. Because even in 0.8 the
> compaction executor is bounded, some of those validations will end up being
> queued behind the first ones. Even assuming that the different validation are
> submitted in the same order on each node (which isn't guaranteed either),
> there is no guarantee that on all nodes, the first validation will take the
> same time, hence desynchronizing the queued ones.
> Overall, it is important for the precision of repair that for a given CF and
> range (which is the unit at which trees are computed), we make sure that all
> node will start the validation at the same time (or, since we can't do magic,
> as close as possible).
> One (reasonably simple) proposition to fix this would be to have repair
> schedule validation compactions across nodes one by one (i.e, one CF/range at
> a time), waiting for all nodes to return their tree before submitting the
> next request. Then on each node, we should make sure that the node will start
> the validation compaction as soon as requested. For that, we probably want to
> have a specific executor for validation compaction and:
> * either we fail the whole repair whenever one node is not able to execute
> the validation compaction right away (because no thread are available right
> away).
> * we simply tell the user that if he start too many repairs in parallel, he
> may start seeing some of those repairing more data than it should.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
