[
https://issues.apache.org/jira/browse/FLINK-34227?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17810745#comment-17810745
]
Matthias Pohl edited comment on FLINK-34227 at 1/25/24 8:18 AM:
----------------------------------------------------------------
*Call Hierarchy*
{code:bash}
JobMaster.tryConnectToResourceManager() (org.apache.flink.runtime.jobmaster)
JobMaster.reconnectToResourceManager(Exception)
(org.apache.flink.runtime.jobmaster)
JobMaster.disconnectResourceManager(ResourceManagerId, Exception)
(org.apache.flink.runtime.jobmaster)
# "INFO: Disconnect job manager [...]" missed
ResourceManager.closeJobManagerConnection(JobID,
ResourceRequirementHandling, Exception)
(org.apache.flink.runtime.resourcemanager)
ResourceManager.removeJob(JobID, Exception)
(org.apache.flink.runtime.resourcemanager)
ResourceManager.disconnectJobManager(JobID, JobStatus,
Exception) (org.apache.flink.runtime.resourcemanager)
# "INFO: Close ResourceManager connection" happened
JobMaster.dissolveResourceManagerConnection(EstablishedResourceManagerConnection,
Exception) (org.apache.flink.runtime.jobmaster)
JobMaster.closeResourceManagerConnection(Exception)
(org.apache.flink.runtime.jobmaster)
JobMaster.reconnectToResourceManager(Exception)
(org.apache.flink.runtime.jobmaster)
JobMaster.disconnectTaskManagerResourceManagerConnections(Exception)
(org.apache.flink.runtime.jobmaster)
JobMaster.stopJobExecution(Exception)
(org.apache.flink.runtime.jobmaster)
JobMaster.onStop()
(org.apache.flink.runtime.jobmaster)
RpcEndpoint.internalCallOnStop()
(org.apache.flink.runtime.rpc)
terminate(PekkoRpcActor<?>,
ClassLoader) in StartedState in PekkoRpcActor
(org.apache.flink.runtime.rpc.pekko)
PekkoRpcActor.handleControlMessage(ControlMessages)
(org.apache.flink.runtime.rpc.pekko)
PekkoRpcActor.createReceive() (org.apache.flink.runtime.rpc.pekko)
{code}
*Findings*
These are reasons I could come up with that might have prevented the disconnect
log message (and as a consequence traversing the call tree further)
* {{ResourceManager#jobManagerRegistrations}} does not include job anymore
([ResourceManager:1082|https://github.com/apache/flink/blob/ab9445aca56e7d139e8fd9bcc23e5f7e06288e66/flink-runtime/src/main/java/org/apache/flink/runtime/resourcemanager/ResourceManager.java#L1082])
** ResourceManager#jobManagerRegistrations is updated in three different
locations:
*** When registering a JobMaster with the ResourceManager
([ResourceManager#registerJobMasterInternal|#registerJobMasterInternal|https://github.com/apache/flink/blob/ab9445aca56e7d139e8fd9bcc23e5f7e06288e66/flink-runtime/src/main/java/org/apache/flink/runtime/resourcemanager/ResourceManager.java#L930])
**** This would be revealed through a log message [INFO: Registering job
manager|https://github.com/apache/flink/blob/ab9445aca56e7d139e8fd9bcc23e5f7e06288e66/flink-runtime/src/main/java/org/apache/flink/runtime/resourcemanager/ResourceManager.java#L388]
*** When clearing the internal state due to a
[ResourceManager#stopResourceManagerServices|#stopResourceManagerServices|https://github.com/apache/flink/blob/ab9445aca56e7d139e8fd9bcc23e5f7e06288e66/flink-runtime/src/main/java/org/apache/flink/runtime/resourcemanager/ResourceManager.java#L350]
**** This only happens if the ResourceManager gets restarted or stopped
*** As part of the JobManagerConnection closing
([ResourceManager#closeJobManagerConnection|#closeJobManagerConnection|https://github.com/apache/flink/blob/ab9445aca56e7d139e8fd9bcc23e5f7e06288e66/flink-runtime/src/main/java/org/apache/flink/runtime/resourcemanager/ResourceManager.java#L1080])
**** This is the case we're in and which should have removed the job actually
but didn't
* Leader change for the ResourceManager cannot be the reason:
** no logs available related to leader election
** the ResourceManager session ID {{891f46bbb398d49ad91e1dde0bee410c}} stays
the same between both test runs
was (Author: mapohl):
*Call Hierarchy*
{code:bash}
JobMaster.tryConnectToResourceManager() (org.apache.flink.runtime.jobmaster)
JobMaster.reconnectToResourceManager(Exception)
(org.apache.flink.runtime.jobmaster)
JobMaster.disconnectResourceManager(ResourceManagerId, Exception)
(org.apache.flink.runtime.jobmaster)
# "INFO: Disconnect job manager [...]" missed
ResourceManager.closeJobManagerConnection(JobID,
ResourceRequirementHandling, Exception)
(org.apache.flink.runtime.resourcemanager)
ResourceManager.removeJob(JobID, Exception)
(org.apache.flink.runtime.resourcemanager)
ResourceManager.disconnectJobManager(JobID, JobStatus,
Exception) (org.apache.flink.runtime.resourcemanager)
# "INFO: Close ResourceManager connection" happened
JobMaster.dissolveResourceManagerConnection(EstablishedResourceManagerConnection,
Exception) (org.apache.flink.runtime.jobmaster)
JobMaster.closeResourceManagerConnection(Exception)
(org.apache.flink.runtime.jobmaster)
JobMaster.reconnectToResourceManager(Exception)
(org.apache.flink.runtime.jobmaster)
JobMaster.disconnectTaskManagerResourceManagerConnections(Exception)
(org.apache.flink.runtime.jobmaster)
JobMaster.stopJobExecution(Exception)
(org.apache.flink.runtime.jobmaster)
JobMaster.onStop()
(org.apache.flink.runtime.jobmaster)
RpcEndpoint.internalCallOnStop()
(org.apache.flink.runtime.rpc)
terminate(PekkoRpcActor<?>,
ClassLoader) in StartedState in PekkoRpcActor
(org.apache.flink.runtime.rpc.pekko)
PekkoRpcActor.handleControlMessage(ControlMessages)
(org.apache.flink.runtime.rpc.pekko)
PekkoRpcActor.createReceive() (org.apache.flink.runtime.rpc.pekko)
{code}
*Findings*
These are reasons I could come up with that might have prevented the disconnect
log message (and as a consequence traversing the call tree further)
* ResourceManager#jobManagerRegistrations does not include job anymore
([ResourceManager:1082|https://github.com/apache/flink/blob/ab9445aca56e7d139e8fd9bcc23e5f7e06288e66/flink-runtime/src/main/java/org/apache/flink/runtime/resourcemanager/ResourceManager.java#L1082])
** ResourceManager#jobManagerRegistrations is updated in three different
locations:
*** When registering a JobMaster with the ResourceManager
([ResourceManager#registerJobMasterInternal|#registerJobMasterInternal|https://github.com/apache/flink/blob/ab9445aca56e7d139e8fd9bcc23e5f7e06288e66/flink-runtime/src/main/java/org/apache/flink/runtime/resourcemanager/ResourceManager.java#L930])
**** This would be revealed through a log message [INFO: Registering job
manager|https://github.com/apache/flink/blob/ab9445aca56e7d139e8fd9bcc23e5f7e06288e66/flink-runtime/src/main/java/org/apache/flink/runtime/resourcemanager/ResourceManager.java#L388]
*** When clearing the internal state due to a
[ResourceManager#stopResourceManagerServices|#stopResourceManagerServices|https://github.com/apache/flink/blob/ab9445aca56e7d139e8fd9bcc23e5f7e06288e66/flink-runtime/src/main/java/org/apache/flink/runtime/resourcemanager/ResourceManager.java#L350]
**** This only happens if the ResourceManager gets restarted or stopped
*** As part of the JobManagerConnection closing
([ResourceManager#closeJobManagerConnection|#closeJobManagerConnection|https://github.com/apache/flink/blob/ab9445aca56e7d139e8fd9bcc23e5f7e06288e66/flink-runtime/src/main/java/org/apache/flink/runtime/resourcemanager/ResourceManager.java#L1080])
**** This is the case we're in and which should have removed the job actually
but didn't
* Leader change for the ResourceManager cannot be the reason:
** no logs available related to leader election
** the ResourceManager session ID {{891f46bbb398d49ad91e1dde0bee410c}} stays
the same between both test runs
> Job doesn't disconnect from ResourceManager
> -------------------------------------------
>
> Key: FLINK-34227
> URL: https://issues.apache.org/jira/browse/FLINK-34227
> Project: Flink
> Issue Type: Bug
> Components: Runtime / Coordination
> Affects Versions: 1.18.1
> Reporter: Matthias Pohl
> Assignee: Matthias Pohl
> Priority: Critical
> Labels: github-actions, test-stability
> Attachments: FLINK-34227.7e7d69daebb438b8d03b7392c9c55115.log,
> FLINK-34227.log
>
>
> https://github.com/XComp/flink/actions/runs/7634987973/job/20800205972#step:10:14557
> {code}
> [...]
> "main" #1 prio=5 os_prio=0 tid=0x00007fcccc4b7000 nid=0x24ec0 waiting on
> condition [0x00007fccce1eb000]
> java.lang.Thread.State: WAITING (parking)
> at sun.misc.Unsafe.park(Native Method)
> - parking to wait for <0x00000000bdd52618> (a
> java.util.concurrent.CompletableFuture$Signaller)
> at java.util.concurrent.locks.LockSupport.park(LockSupport.java:175)
> at
> java.util.concurrent.CompletableFuture$Signaller.block(CompletableFuture.java:1707)
> at
> java.util.concurrent.ForkJoinPool.managedBlock(ForkJoinPool.java:3323)
> at
> java.util.concurrent.CompletableFuture.waitingGet(CompletableFuture.java:1742)
> at
> java.util.concurrent.CompletableFuture.get(CompletableFuture.java:1908)
> at
> org.apache.flink.streaming.api.environment.StreamExecutionEnvironment.execute(StreamExecutionEnvironment.java:2131)
> at
> org.apache.flink.streaming.api.environment.StreamExecutionEnvironment.execute(StreamExecutionEnvironment.java:2099)
> at
> org.apache.flink.streaming.api.environment.StreamExecutionEnvironment.execute(StreamExecutionEnvironment.java:2077)
> at
> org.apache.flink.streaming.api.scala.StreamExecutionEnvironment.execute(StreamExecutionEnvironment.scala:876)
> at
> org.apache.flink.table.planner.runtime.stream.sql.WindowDistinctAggregateITCase.testHopWindow_Cube(WindowDistinctAggregateITCase.scala:550)
> [...]
> {code}
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
