[jira] [Created] (FLINK-12319) StackOverFlowError in cep.nfa.sharedbuffer.SharedBuffer

[Shang Yuanchun (Jira)]

Marco Pfatschbacher created FLINK-12319:
-------------------------------------------

             Summary: StackOverFlowError in cep.nfa.sharedbuffer.SharedBuffer
                 Key: FLINK-12319
                 URL: https://issues.apache.org/jira/browse/FLINK-12319
             Project: Flink
          Issue Type: Bug
          Components: Library / CEP
    Affects Versions: 1.8.0, 1.7.2, 1.6.4
         Environment: Ubuntu 18.04

openjdk version "1.8.0_191"
OpenJDK Runtime Environment (build 1.8.0_191-8u191-b12-2ubuntu0.18.04.1-b12)
OpenJDK 64-Bit Server VM (build 25.191-b12, mixed mode)
            Reporter: Marco Pfatschbacher


 

I wrote a simple SourceFunction that creats Events in a loop.

The CEP pattern is very simple:
{code:java}
      final Pattern<Event, ?> failurePattern =
              Pattern.<Event>begin("5 or more failures", AfterMatchSkipStrategy.skipPastLastEvent())
                      .subtype(LoginEvent.class)
                      .where(
                              new IterativeCondition<LoginEvent>() {
                                  @Override
                                  public boolean filter(LoginEvent value, Context<LoginEvent> ctx) throws Exception {
                                      return value.get("type").equals("failed");
                                  }
                              })
                      .times(5)
                      .next("1 or more successes")
                      .subtype(LoginEvent.class)
                      .where(
                              new IterativeCondition<LoginEvent>() {
                                  @Override
                                  public boolean filter(LoginEvent value, Context<LoginEvent> ctx) throws Exception {
                                      return value.get("type").equals("success");
                                  }
                              })
                      .times(1)
                      .within(Time.milliseconds(20));
{code}
 

After about 100k Events, Flink aborts with this stacktrace:

 
{noformat}
Exception in thread "main" org.apache.flink.runtime.client.JobExecutionException: Job execution failed.
    at org.apache.flink.runtime.jobmaster.JobResult.toJobExecutionResult(JobResult.java:146)
    at org.apache.flink.runtime.minicluster.MiniCluster.executeJobBlocking(MiniCluster.java:630)
    at org.apache.flink.streaming.api.environment.LocalStreamEnvironment.execute(LocalStreamEnvironment.java:123)
    at org.classdump.alerts.FlinkCep.brute_force_login(FlinkCep.java:263)
    at org.classdump.alerts.FlinkCep.main(FlinkCep.java:41)
Caused by: java.lang.StackOverflowError
    at org.apache.flink.runtime.state.heap.HeapMapState.get(HeapMapState.java:85)
    at org.apache.flink.runtime.state.UserFacingMapState.get(UserFacingMapState.java:47)
    at org.apache.flink.cep.nfa.sharedbuffer.SharedBuffer.releaseNode(SharedBuffer.java:339)
    at org.apache.flink.cep.nfa.sharedbuffer.SharedBuffer.removeNode(SharedBuffer.java:355)
    at org.apache.flink.cep.nfa.sharedbuffer.SharedBuffer.releaseNode(SharedBuffer.java:342)
    at org.apache.flink.cep.nfa.sharedbuffer.SharedBuffer.removeNode(SharedBuffer.java:355)
    at org.apache.flink.cep.nfa.sharedbuffer.SharedBuffer.releaseNode(SharedBuffer.java:342)
    at org.apache.flink.cep.nfa.sharedbuffer.SharedBuffer.removeNode(SharedBuffer.java:355)
    at org.apache.flink.cep.nfa.sharedbuffer.SharedBuffer.releaseNode(SharedBuffer.java:342)
    at org.apache.flink.cep.nfa.sharedbuffer.SharedBuffer.removeNode(SharedBuffer.java:355)
    at org.apache.flink.cep.nfa.sharedbuffer.SharedBuffer.releaseNode(SharedBuffer.java:342)
    at org.apache.flink.cep.nfa.sharedbuffer.SharedBuffer.removeNode(SharedBuffer.java:355)
    at org.apache.flink.cep.nfa.sharedbuffer.SharedBuffer.releaseNode(SharedBuffer.java:342)
    at org.apache.flink.cep.nfa.sharedbuffer.SharedBuffer.removeNode(SharedBuffer.java:355)
    at org.apache.flink.cep.nfa.sharedbuffer.SharedBuffer.releaseNode(SharedBuffer.java:342)
    at org.apache.flink.cep.nfa.sharedbuffer.SharedBuffer.removeNode(SharedBuffer.java:355)
    at org.apache.flink.cep.nfa.sharedbuffer.SharedBuffer.releaseNode(SharedBuffer.java:342)
    at org.apache.flink.cep.nfa.sharedbuffer.SharedBuffer.removeNode(SharedBuffer.java:355)
    at org.apache.flink.cep.nfa.sharedbuffer.SharedBuffer.releaseNode(SharedBuffer.java:342)
    at org.apache.flink.cep.nfa.sharedbuffer.SharedBuffer.removeNode(SharedBuffer.java:355)
    at org.apache.flink.cep.nfa.sharedbuffer.SharedBuffer.releaseNode(SharedBuffer.java:342)
    at org.apache.flink.cep.nfa.sharedbuffer.SharedBuffer.removeNode(SharedBuffer.java:355){noformat}
 

This happens with version 1.8.0, 1.7.2, 1.6.4

Version 1.5.6 does not have this issue.

Seems to be related to FLINK-9418

 

 



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)