[
https://issues.apache.org/jira/browse/CASSANDRA-15902?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17224837#comment-17224837
]
Swen Fuhrmann commented on CASSANDRA-15902:
-------------------------------------------
[~adejanovski] thanks for the review, [~brandon.williams] thanks for committing!
[~mck] Not sure if fixed for version "4.0" and "4.0-beta3" correct for this
ticket. As far as I see there was not merge in trunk. This issue only appear in
3.x (see my comment above). I only created a regression test for trunk (but was
not committed yet).
Should we add the test to trunk?
Should we also merge the fix to trunk because improves the code even thought
its not an issue any longer?
I'd be glad to prepare this fix also for trunk if that makes sense.
Any thoughts?
> OOM because repair session thread not closed when terminating repair
> --------------------------------------------------------------------
>
> Key: CASSANDRA-15902
> URL: https://issues.apache.org/jira/browse/CASSANDRA-15902
> Project: Cassandra
> Issue Type: Bug
> Components: Consistency/Repair
> Reporter: Swen Fuhrmann
> Assignee: Swen Fuhrmann
> Priority: Normal
> Fix For: 4.0, 3.0.23, 3.11.9, 4.0-beta3
>
> Attachments: heap-mem-histo.txt, repair-terminated.txt
>
>
> In our cluster, after a while some nodes running slowly out of memory. On
> that nodes we observed that Cassandra Reaper terminate repairs with a JMX
> call to {{StorageServiceMBean.forceTerminateAllRepairSessions()}} because
> reaching timeout of 30 min.
> In the memory heap dump we see lot of instances of
> {{io.netty.util.concurrent.FastThreadLocalThread}} occupy most of the memory:
> {noformat}
> 119 instances of "io.netty.util.concurrent.FastThreadLocalThread", loaded by
> "sun.misc.Launcher$AppClassLoader @ 0x51a800000" occupy 8.445.684.480 (93,96
> %) bytes. {noformat}
> In the thread dump we see lot of repair threads:
> {noformat}
> grep "Repair#" threaddump.txt | wc -l
> 50 {noformat}
>
> The repair jobs are waiting for the validation to finish:
> {noformat}
> "Repair#152:1" #96170 daemon prio=5 os_prio=0 tid=0x0000000012fc5000
> nid=0x542a waiting on condition [0x00007f81ee414000]
> java.lang.Thread.State: WAITING (parking)
> at sun.misc.Unsafe.park(Native Method)
> - parking to wait for <0x00000007939bcfc8> (a
> com.google.common.util.concurrent.AbstractFuture$Sync)
> at java.util.concurrent.locks.LockSupport.park(LockSupport.java:175)
> at
> java.util.concurrent.locks.AbstractQueuedSynchronizer.parkAndCheckInterrupt(AbstractQueuedSynchronizer.java:836)
> at
> java.util.concurrent.locks.AbstractQueuedSynchronizer.doAcquireSharedInterruptibly(AbstractQueuedSynchronizer.java:997)
> at
> java.util.concurrent.locks.AbstractQueuedSynchronizer.acquireSharedInterruptibly(AbstractQueuedSynchronizer.java:1304)
> at
> com.google.common.util.concurrent.AbstractFuture$Sync.get(AbstractFuture.java:285)
> at
> com.google.common.util.concurrent.AbstractFuture.get(AbstractFuture.java:116)
> at
> com.google.common.util.concurrent.Uninterruptibles.getUninterruptibly(Uninterruptibles.java:137)
> at
> com.google.common.util.concurrent.Futures.getUnchecked(Futures.java:1509)
> at org.apache.cassandra.repair.RepairJob.run(RepairJob.java:160)
> at
> java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149)
> at
> java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)
> at
> org.apache.cassandra.concurrent.NamedThreadFactory.lambda$threadLocalDeallocator$0(NamedThreadFactory.java:81)
> at
> org.apache.cassandra.concurrent.NamedThreadFactory$$Lambda$13/480490520.run(Unknown
> Source)
> at java.lang.Thread.run(Thread.java:748) {noformat}
>
> Thats the line where the threads stuck:
> {noformat}
> // Wait for validation to complete
> Futures.getUnchecked(validations); {noformat}
>
> The call to {{StorageServiceMBean.forceTerminateAllRepairSessions()}} stops
> the thread pool executor. It looks like that futures which are in progress
> will therefor never be completed and the repair thread waits forever and
> won't be finished.
>
> Environment:
> Cassandra version: 3.11.4 and 3.11.6
> Cassandra Reaper: 1.4.0
> JVM memory settings:
> {noformat}
> -Xms11771M -Xmx11771M -XX:+UseG1GC -XX:MaxGCPauseMillis=100
> -XX:+ParallelRefProcEnabled -XX:MaxMetaspaceSize=100M {noformat}
> on another cluster with same issue:
> {noformat}
> -Xms31744M -Xmx31744M -XX:+UseG1GC -XX:MaxGCPauseMillis=100
> -XX:+ParallelRefProcEnabled -XX:MaxMetaspaceSize=100M {noformat}
> Java Runtime:
> {noformat}
> openjdk version "1.8.0_212"
> OpenJDK Runtime Environment (AdoptOpenJDK)(build 1.8.0_212-b03)
> OpenJDK 64-Bit Server VM (AdoptOpenJDK)(build 25.212-b03, mixed mode)
> {noformat}
>
> The same issue described in this comment:
> https://issues.apache.org/jira/browse/CASSANDRA-14355?focusedCommentId=16992973&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-16992973
> As suggested in the comments I created this new specific ticket.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
