cassandra-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "T Jake Luciani (JIRA)" <>
Subject [jira] [Commented] (CASSANDRA-4492) HintsColumnFamily compactions hang when using multithreaded compaction
Date Tue, 16 Oct 2012 18:03:03 GMT


T Jake Luciani commented on CASSANDRA-4492:

I hit this recently as well.

Though mine I was able to reproduce.  If you call truncate while a compaction is currently
going on it hangs both the truncate and the parallel compaction iterator.  

> HintsColumnFamily compactions hang when using multithreaded compaction
> ----------------------------------------------------------------------
>                 Key: CASSANDRA-4492
>                 URL:
>             Project: Cassandra
>          Issue Type: Bug
>    Affects Versions: 1.0.11
>            Reporter: Jason Harvey
>            Priority: Minor
>         Attachments: jstack.txt
> Running into an issue on a 6 node ring running 1.0.11 where HintsColumnFamily compactions
often hang indefinitely when using multithreaded compaction. Nothing of note in the logs.
In some cases, the compaction hangs before a tmp sstable is even created.
> I've wiped out every hints sstable and restarted several times. The issue always comes
back rather quickly and predictably. The compactions sometimes complete if the hint sstables
are very small. Disabling multithreaded compaction stops this issue from occurring.
> Compactions of all other CFs seem to work just fine.
> This ring was upgraded from 1.0.7. I didn't keep any hints from the upgrade.
> I should note that the ring gets a huge amount of writes, and as a result the HintedHandoff
rows get be quite wide. I didn't see any large-row compaction notices when the compaction
was hanging (perhaps the bug was triggered by incremental compaction?). After disabling multithreaded
compaction, several of the rows that were successfully compacted were over 1GB.
> Here is the output I see from compactionstats where a compaction has hung. The 'bytes
compacted' column never changes.
> {code}
> pending tasks: 1
>           compaction type        keyspace   column family bytes compacted     bytes total
>                Compaction          systemHintsColumnFamily          268082       464784758
> {code}
> The hung thread stack is as follows: (full jstack attached, as well)
> {code}
> "CompactionExecutor:37" daemon prio=10 tid=0x00000000063df800 nid=0x49d9 waiting on condition
>    java.lang.Thread.State: WAITING (parking)
>         at sun.misc.Unsafe.park(Native Method)
>         - parking to wait for  <0x000000050f2e0e58> (a java.util.concurrent.locks.AbstractQueuedSynchronizer$ConditionObject)
>         at java.util.concurrent.locks.LockSupport.park(
>         at java.util.concurrent.locks.AbstractQueuedSynchronizer$ConditionObject.await(
>         at java.util.concurrent.LinkedBlockingQueue.take(
>         at org.apache.cassandra.db.compaction.ParallelCompactionIterable$Deserializer.computeNext(
>         at org.apache.cassandra.db.compaction.ParallelCompactionIterable$Deserializer.computeNext(
>         at
>         at
>         at org.apache.cassandra.utils.MergeIterator$Candidate.advance(
>         at org.apache.cassandra.utils.MergeIterator$ManyToOne.advance(
>         at org.apache.cassandra.utils.MergeIterator$ManyToOne.computeNext(
>         at
>         at
>         at org.apache.cassandra.db.compaction.ParallelCompactionIterable$Unwrapper.computeNext(
>         at org.apache.cassandra.db.compaction.ParallelCompactionIterable$Unwrapper.computeNext(
>         at
>         at
>         at$7.computeNext(
>         at
>         at
>         at org.apache.cassandra.db.compaction.CompactionTask.execute(
>         at org.apache.cassandra.db.compaction.CompactionManager$
>         at java.util.concurrent.FutureTask$Sync.innerRun(
>         at
>         at java.util.concurrent.ThreadPoolExecutor$Worker.runTask(
>         at java.util.concurrent.ThreadPoolExecutor$
>         at
> {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:

View raw message