[
https://issues.apache.org/jira/browse/IGNITE-20160?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Sergey Uttsel resolved IGNITE-20160.
------------------------------------
Resolution: Duplicate
Fixed in https://issues.apache.org/jira/browse/IGNITE-20774
> NullPointerException in FSMCallerImpl.doCommitted
> -------------------------------------------------
>
> Key: IGNITE-20160
> URL: https://issues.apache.org/jira/browse/IGNITE-20160
> Project: Ignite
> Issue Type: Bug
> Affects Versions: 3.0.0-beta1
> Reporter: Pavel Tupitsyn
> Assignee: Sergey Uttsel
> Priority: Blocker
> Labels: ignite-3
> Fix For: 3.0.0-beta2
>
> Time Spent: 20m
> Remaining Estimate: 0h
>
> {code:java}
> java.lang.NullPointerException
> at
> org.apache.ignite.raft.jraft.core.FSMCallerImpl.doCommitted(FSMCallerImpl.java:496)
> at
> org.apache.ignite.raft.jraft.core.FSMCallerImpl.runApplyTask(FSMCallerImpl.java:448)
> at
> org.apache.ignite.raft.jraft.core.FSMCallerImpl$ApplyTaskHandler.onEvent(FSMCallerImpl.java:136)
> at
> org.apache.ignite.raft.jraft.core.FSMCallerImpl$ApplyTaskHandler.onEvent(FSMCallerImpl.java:130)
> at
> org.apache.ignite.raft.jraft.disruptor.StripedDisruptor$StripeEntryHandler.onEvent(StripedDisruptor.java:226)
> at
> org.apache.ignite.raft.jraft.disruptor.StripedDisruptor$StripeEntryHandler.onEvent(StripedDisruptor.java:191)
> at
> com.lmax.disruptor.BatchEventProcessor.run(BatchEventProcessor.java:137)
> {code}
> [https://ci.ignite.apache.org/buildConfiguration/ApacheIgnite3xGradle_Test_IntegrationTests_ModuleRunnerSqlLogic/7410174?hideProblemsFromDependencies=false&hideTestsFromDependencies=false&expandBuildProblemsSection=true&expandBuildChangesSection=true]
> It happens here (see FSMCallerImpl#doCommitted):
> {code:java}
> final IteratorImpl iterImpl = new IteratorImpl(this.fsm, this.logManager,
> closures, firstClosureIndex,
> lastAppliedIndex, committedIndex, this.applyingIndex,
> this.node.getOptions());{code}
> on the 2nd line, most likely on resolving null pointer to *node,* which is
> nullified on FSMCaller shutdown. Raft groups were being stopped in this
> moment.
> *Implementation details*
> A simple fix to avoid the NPE at the aforementioned line would be to check
> `node` for null.
> Additionally it would be nice to check `shutdownLatch` in `doCommitted` and
> make sure we call `unsubscribe` in a proper order.
> The reason is that doCommitted is called from a disruptor's callback.
> One more observation - the reference to a node is set to null in `shutdown`
> that is called before `join` where we unsubscribe from notifications from
> Disruptor. There is a little chance that something comes into FSMCallerImpl
> after shutdown but before join.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
