[
https://issues.apache.org/jira/browse/SAMZA-347?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14484283#comment-14484283
]
Chris Riccomini commented on SAMZA-347:
---------------------------------------
bq. Do you know why this happens
I suspect it's because you're running the same test multiple times. Without a
close call, I think the test stops with the RocksDB connection still open (the
LOCK file is still there). If you then run the test again, it will still see
the LOCK file, and give this. I don't think it's a big deal.
Thanks, I'll close the ticket!
> Failure in SamzaContainer causes LevelDB stack trace
> ----------------------------------------------------
>
> Key: SAMZA-347
> URL: https://issues.apache.org/jira/browse/SAMZA-347
> Project: Samza
> Issue Type: Bug
> Components: kv
> Affects Versions: 0.9.0
> Reporter: Chris Riccomini
> Assignee: Renato Javier Marroquín Mogrovejo
> Labels: newbie
> Fix For: 0.10.0
>
> Attachments: SAMZA-347.patch
>
>
> If there is a failure in SamzaContainer, and state is configured with
> LevelDB, we see this exception:
> {noformat}
> Exception in thread "main"
> org.fusesource.leveldbjni.internal.NativeDB$DBException: IO error:
> /mnt/u002/samsa-yarn/usercache/samza-perf-playground/appcache/application_1399672625491_0002/container_1399672625491_0002_01_000611/state/test-store-all-calls/47/LOCK:
> No such file or directory
> at
> org.fusesource.leveldbjni.internal.NativeDB.checkStatus(NativeDB.java:200)
> at org.fusesource.leveldbjni.internal.NativeDB.open(NativeDB.java:218)
> at org.fusesource.leveldbjni.JniDBFactory.open(JniDBFactory.java:168)
> at
> org.apache.samza.storage.kv.LevelDbKeyValueStore.db$lzycompute(LevelDbKeyValueStore.scala:71)
> at
> org.apache.samza.storage.kv.LevelDbKeyValueStore.db(LevelDbKeyValueStore.scala:71)
> at
> org.apache.samza.storage.kv.LevelDbKeyValueStore.putAll(LevelDbKeyValueStore.scala:99)
> at
> org.apache.samza.storage.kv.SerializedKeyValueStore.putAll(SerializedKeyValueStore.scala:58)
> at org.apache.samza.storage.kv.CachedStore.flush(CachedStore.scala:161)
> at
> org.apache.samza.storage.kv.NullSafeKeyValueStore.flush(NullSafeKeyValueStore.scala:68)
> at
> org.apache.samza.storage.kv.KeyValueStorageEngine.flush(KeyValueStorageEngine.scala:117)
> at
> org.apache.samza.storage.kv.KeyValueStorageEngine.close(KeyValueStorageEngine.scala:129)
> at
> org.apache.samza.storage.kv.KeyValueStorageEngine.stop(KeyValueStorageEngine.scala:123)
> at
> org.apache.samza.storage.TaskStorageManager$$anonfun$stop$2.apply(TaskStorageManager.scala:129)
> at
> org.apache.samza.storage.TaskStorageManager$$anonfun$stop$2.apply(TaskStorageManager.scala:129)
> at scala.collection.Iterator$class.foreach(Iterator.scala:727)
> at scala.collection.AbstractIterator.foreach(Iterator.scala:1157)
> at
> scala.collection.MapLike$DefaultValuesIterable.foreach(MapLike.scala:206)
> at
> org.apache.samza.storage.TaskStorageManager.stop(TaskStorageManager.scala:129)
> at
> org.apache.samza.container.TaskInstance.shutdownStores(TaskInstance.scala:236)
> at
> org.apache.samza.container.SamzaContainer$$anonfun$shutdownStores$2.apply(SamzaContainer.scala:660)
> at
> org.apache.samza.container.SamzaContainer$$anonfun$shutdownStores$2.apply(SamzaContainer.scala:660)
> at scala.collection.Iterator$class.foreach(Iterator.scala:727)
> at scala.collection.AbstractIterator.foreach(Iterator.scala:1157)
> at
> scala.collection.MapLike$DefaultValuesIterable.foreach(MapLike.scala:206)
> at
> org.apache.samza.container.SamzaContainer.shutdownStores(SamzaContainer.scala:660)
> at
> org.apache.samza.container.SamzaContainer.run(SamzaContainer.scala:524)
> at
> org.apache.samza.container.SamzaContainer$.main(SamzaContainer.scala:82)
> at org.apache.samza.container.SamzaContainer.main(SamzaContainer.scala)
> {noformat}
> I think that this is because SamzaContainer catches the thrown exception, and
> then calls stop on everything. It seems that calling stop on a
> not-fully-initialized LevelDB JNI store can throw this exception. We should
> either wrap LevelDB's stop logic in a try/catch, or change the logic to check
> that shutting down LevelDB is safe.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
