Details
-
Bug
-
Status: Open
-
Normal
-
Resolution: Unresolved
-
None
-
None
-
CentOS 7.3
-
Normal
Description
We have 14 nodes cluster where we seen hints file getting corrupted and resulting in the following error
ERROR [HintsDispatcher:1] 2018-04-06 16:26:44,423 CassandraDaemon.java:228 - Exception in thread Thread[HintsDispatcher:1,1,main] org.apache.cassandra.io.FSReadError: java.io.IOException: Digest mismatch exception at org.apache.cassandra.hints.HintsReader$BuffersIterator.computeNext(HintsReader.java:298) ~[apache-cassandra-3.11.1.jar:3.11.1-SNAPSHOT] at org.apache.cassandra.hints.HintsReader$BuffersIterator.computeNext(HintsReader.java:263) ~[apache-cassandra-3.11.1.jar:3.11.1-SNAPSHOT] at org.apache.cassandra.utils.AbstractIterator.hasNext(AbstractIterator.java:47) ~[apache-cassandra-3.11.1.jar:3.11.1-SNAPSHOT] at org.apache.cassandra.hints.HintsDispatcher.sendHints(HintsDispatcher.java:169) ~[apache-cassandra-3.11.1.jar:3.11.1-SNAPSHOT] at org.apache.cassandra.hints.HintsDispatcher.sendHintsAndAwait(HintsDispatcher.java:128) ~[apache-cassandra-3.11.1.jar:3.11.1-SNAPSHOT] at org.apache.cassandra.hints.HintsDispatcher.dispatch(HintsDispatcher.java:113) ~[apache-cassandra-3.11.1.jar:3.11.1-SNAPSHOT] at org.apache.cassandra.hints.HintsDispatcher.dispatch(HintsDispatcher.java:94) ~[apache-cassandra-3.11.1.jar:3.11.1-SNAPSHOT] at org.apache.cassandra.hints.HintsDispatchExecutor$DispatchHintsTask.deliver(HintsDispatchExecutor.java:278) ~[apache-cassandra-3.11.1.jar:3.11.1-SNAPSHOT] at org.apache.cassandra.hints.HintsDispatchExecutor$DispatchHintsTask.dispatch(HintsDispatchExecutor.java:260) ~[apache-cassandra-3.11.1.jar:3.11.1-SNAPSHOT] at org.apache.cassandra.hints.HintsDispatchExecutor$DispatchHintsTask.dispatch(HintsDispatchExecutor.java:238) ~[apache-cassandra-3.11.1.jar:3.11.1-SNAPSHOT] at org.apache.cassandra.hints.HintsDispatchExecutor$DispatchHintsTask.run(HintsDispatchExecutor.java:217) ~[apache-cassandra-3.11.1.jar:3.11.1-SNAPSHOT] at java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:511) ~[na:1.8.0_141] at java.util.concurrent.FutureTask.run(FutureTask.java:266) ~[na:1.8.0_141] at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149) ~[na:1.8.0_141] at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624) [na:1.8.0_141] at org.apache.cassandra.concurrent.NamedThreadFactory.lambda$threadLocalDeallocator$0(NamedThreadFactory.java:81) [apache-cassandra-3.11.1.jar:3.11.1-SNAPSHOT] at java.lang.Thread.run(Thread.java:748) ~[na:1.8.0_141] Caused by: java.io.IOException: Digest mismatch exception at org.apache.cassandra.hints.HintsReader$BuffersIterator.computeNextInternal(HintsReader.java:315) ~[apache-cassandra-3.11.1.jar:3.11.1-SNAPSHOT] at org.apache.cassandra.hints.HintsReader$BuffersIterator.computeNext(HintsReader.java:289) ~[apache-cassandra-3.11.1.jar:3.11.1-SNAPSHOT] ... 16 common frames omitted
Notes on cluster and investigation done so far
1. Cassandra used here is built locally from 3.11.1 branch along with following patch from issue: CASSANDRA-14080
https://github.com/apache/cassandra/commit/68079e4b2ed4e58dbede70af45414b3d4214e195
2. The bootstrap of 14 nodes happens in the following way:
- Out of 14 nodes only 3 nodes are picked as seed nodes.
- Only 1 out 3 seed nodes is started and schema is created if it was not created previously.
- Post this, rest of nodes are bootstrapped.
- In failure scenario, only 5 out of 14 succesfully formed the cassandra cluster. The failed nodes include two seed nodes.
3. We confirmed the following patch from issue:CASSANDRA-13696has been applied. From confirmed from Jay Zhuang that this is different issue from what was previously fixed.
"this should be a different issue, as HintsDispatcher.java:128 sends hints with {{buffer}}s, this patch is only to fix the digest mismatch for HintsDispatcher.java:129, which sends hints one by one."
4. Application uses java driver with quoram setting for cassandra
5. We saw this issue on 7 node cluster too (different from 14 node cluster)
6. We are able to workaround by running nodetool truncatehints on failed nodes and restarting cassandra.