[
https://issues.apache.org/jira/browse/CASSANDRA-5146?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13555230#comment-13555230
]
Sylvain Lebresne commented on CASSANDRA-5146:
---------------------------------------------
bq. why would a node be receiving a merkle tree from itself?
Because we didn't bothered "specializing" the local case. So the repair
coordinator sends itself a merkle tree request as any other node and later
sends itself the tree oblivious to the fact it is itself the coordinator of
said repair. It's definitively not very efficient, as we serialize/deserialize
the tree uselessly, but on the other side it's not in any performance critical
path. If it bugs your OCD (no judgment, we're an OCD-friendly project), feel
free to open a ticket for the improvement (and maybe giving a shot at a patch?).
For the hanging problem, all the log above is saying is that 10.8.30.14 hasn't
responded with his merkle tree. Maybe check the log of said node around the
time the tree request was sent and check if something jumps out. Is that
something you can reproduce? If it is, maybe reproducing with DEBUG logging on
might shed some light.
> repair -pr hangs
> ----------------
>
> Key: CASSANDRA-5146
> URL: https://issues.apache.org/jira/browse/CASSANDRA-5146
> Project: Cassandra
> Issue Type: Bug
> Components: Core
> Affects Versions: 1.2.0
> Environment: Ubuntu 12.04
> Reporter: Michael Kjellman
> Assignee: Sylvain Lebresne
> Priority: Critical
>
> while running a repair -pr the repair seems to hang after getting a merkle
> tree
> {code}
> INFO [AntiEntropySessions:9] 2013-01-10 18:23:01,652 AntiEntropyService.java
> (line 652) [repair #d29fd100-5b95-11e2-b9c7-dd50a26832ff] new session: will
> sync /10.8.25.101, /10.8.30.14 on range
> (28356863910078205288614550619314017620,42535295865117307932921825928971026436]
> for evidence.[fingerprints, messages]
> INFO [AntiEntropySessions:9] 2013-01-10 18:23:01,653 AntiEntropyService.java
> (line 857) [repair #d29fd100-5b95-11e2-b9c7-dd50a26832ff] requesting merkle
> trees for fingerprints (to [/10.8.30.14, /10.8.25.101])
> INFO [ValidationExecutor:7] 2013-01-10 18:23:01,654 ColumnFamilyStore.java
> (line 647) Enqueuing flush of
> Memtable-fingerprints@500862962(12960712/12960712 serialized/live bytes, 469
> ops)
> INFO [FlushWriter:25] 2013-01-10 18:23:01,655 Memtable.java (line 424)
> Writing Memtable-fingerprints@500862962(12960712/12960712 serialized/live
> bytes, 469 ops)
> INFO [FlushWriter:25] 2013-01-10 18:23:02,058 Memtable.java (line 458)
> Completed flushing
> /data2/cassandra/evidence/fingerprints/evidence-fingerprints-ib-192-Data.db
> (11413718 bytes) for commitlog position
> ReplayPosition(segmentId=1357767160463, position=8921654)
> INFO [AntiEntropyStage:1] 2013-01-10 18:25:52,735 AntiEntropyService.java
> (line 214) [repair #d29fd100-5b95-11e2-b9c7-dd50a26832ff] Received merkle
> tree for fingerprints from /10.8.25.101
> {code}
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
