You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@geode.apache.org by "Darrel Schneider (JIRA)" <ji...@apache.org> on 2016/12/21 18:51:58 UTC
[jira] [Commented] (GEODE-2240) unexpected NullPointerException
from Tombstone service
[ https://issues.apache.org/jira/browse/GEODE-2240?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15767845#comment-15767845 ]
Darrel Schneider commented on GEODE-2240:
-----------------------------------------
I think this bug might be caused by the code that adds to "expiredTombstones", TombstoneService.ReplicateTombstoneSweeper.expireTombstone, not syncing on getBlockGCLock() as the code that removes from "expiredTomstones", TombstoneService.ReplicateTombstoneSweeper.removeExpiredIf, does.
Since these are the only two methods that modify the "expiredTombstones" collection they need a common sync to prevent unsafe concurrent modification of the ArrayList.
I think the only time multiple threads would do this concurrently is when a region is being destroyed.
> unexpected NullPointerException from Tombstone service
> ------------------------------------------------------
>
> Key: GEODE-2240
> URL: https://issues.apache.org/jira/browse/GEODE-2240
> Project: Geode
> Issue Type: Bug
> Components: regions
> Reporter: Darrel Schneider
>
> A test failed and the logs were found to be full of NPEs from the tombstone service:[severe 2016/12/20 02:04:35.605 UTC dataStoregemfire7_rs-StorageBTTest-2016-12-19-23-35-42-client-14_19508 <Replicate/Partition Region Garbage Collector> tid=0x44] GemFire garbage collection service encountered an unexpected exception
> java.lang.NullPointerException
> at org.apache.geode.internal.cache.TombstoneService$TombstoneSweeper.lambda$purgeObsoleteTombstones$1(TombstoneService.java:938)
> at org.apache.geode.internal.cache.TombstoneService$ReplicateTombstoneSweeper.removeExpiredIf(TombstoneService.java:479)
> at org.apache.geode.internal.cache.TombstoneService$TombstoneSweeper.removeIf(TombstoneService.java:823)
> at org.apache.geode.internal.cache.TombstoneService$TombstoneSweeper.purgeObsoleteTombstones(TombstoneService.java:937)
> at org.apache.geode.internal.cache.TombstoneService$TombstoneSweeper.run(TombstoneService.java:880)
> at java.lang.Thread.run(Thread.java:745)
> [severe 2016/12/20 02:05:45.987 UTC dataStoregemfire7_rs-StorageBTTest-2016-12-19-23-35-42-client-14_19508 <Replicate/Partition Region Garbage Collector> tid=0x44] GemFire garbage collection service encountered an unexpected exception
> java.lang.NullPointerException
> at org.apache.geode.internal.cache.TombstoneService$ReplicateTombstoneSweeper.expireBatch(TombstoneService.java:524)
> at org.apache.geode.internal.cache.TombstoneService$ReplicateTombstoneSweeper.checkExpiredTombstoneGC(TombstoneService.java:594)
> at org.apache.geode.internal.cache.TombstoneService$TombstoneSweeper.run(TombstoneService.java:878)
> at java.lang.Thread.run(Thread.java:745)
> Both of these stacks indicate that the "expiredTombstones" ArrayList somehow has nulls in it. It is an ArrayList of Tombstone instances and the only code that adds to it first tests that the item it is adding is not null. The only other modify operation done on it is to remove an item.
> Perhaps unsafe concurrent access is happening causing this code to see nulls.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)