[
https://issues.apache.org/jira/browse/HIVE-18547?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Sergey Shelukhin updated HIVE-18547:
------------------------------------
Description:
https://builds.apache.org/job/PreCommit-HIVE-Build/8818/testReport/org.apache.hive.jdbc/TestTriggersMoveWorkloadManager/testTriggerMoveAndKill/
Looks like the cluster allocation assignment and WM event creation race,
probably because WM returns session to the caller ASAP and then makes the
changes after that.
{noformat}
'Event: GET Pool: BI Cluster %: 80.00' expected in STDERR capture, but not
found.
...
2018-01-24T15:07:31,746 INFO [Workload management master] tez.WorkloadManager:
Processing changes for pool BI: [BI, query parallelism 1, fraction of the
cluster 0.800000011920929, fraction used by child pools 0.0, active sessions 0,
initializing sessions 0]
2018-01-24T15:07:31,746 INFO [Workload management master] tez.WorkloadManager:
Starting 1 queries in pool [BI, query parallelism 1, fraction of the cluster
0.800000011920929, fraction used by child pools 0.0, active sessions 0,
initializing sessions 0]
2018-01-24T15:07:31,746 INFO [Workload management master] tez.WorkloadManager:
Received a session from AM pool sessionId=2be29c62-9f2c-40b7-a5eb-6298baf83a34,
queueName=default, user=hiveptest, doAs=false, isOpen=true, isDefault=true,
expires in 588529859ms, WM state poolName=null, clusterFraction=0.0,
queryId=null, killReason=null
2018-01-24T15:07:31,746 INFO [HiveServer2-Background-Pool: Thread-1377]
tez.WmEvent: Added WMEvent: EventType: GET EventStartTimestamp: 1516835251746
elapsedTime: 0 wmTezSessionInfo:SessionId: 2be29c62-9f2c-40b7-a5eb-6298baf83a34
Pool: BI Cluster %: 0.0
2018-01-24T15:07:31,746 INFO [Workload management master]
tez.GuaranteedTasksAllocator: Updating 2be29c62-9f2c-40b7-a5eb-6298baf83a34
with 3 guaranteed tasks
{noformat}
was:
https://builds.apache.org/job/PreCommit-HIVE-Build/8818/testReport/org.apache.hive.jdbc/TestTriggersMoveWorkloadManager/testTriggerMoveAndKill/
Looks like the cluster allocation assignment and WM event creation race somehow
{noformat}
'Event: GET Pool: BI Cluster %: 80.00' expected in STDERR capture, but not
found.
...
2018-01-24T15:07:31,746 INFO [Workload management master] tez.WorkloadManager:
Processing changes for pool BI: [BI, query parallelism 1, fraction of the
cluster 0.800000011920929, fraction used by child pools 0.0, active sessions 0,
initializing sessions 0]
2018-01-24T15:07:31,746 INFO [Workload management master] tez.WorkloadManager:
Starting 1 queries in pool [BI, query parallelism 1, fraction of the cluster
0.800000011920929, fraction used by child pools 0.0, active sessions 0,
initializing sessions 0]
2018-01-24T15:07:31,746 INFO [Workload management master] tez.WorkloadManager:
Received a session from AM pool sessionId=2be29c62-9f2c-40b7-a5eb-6298baf83a34,
queueName=default, user=hiveptest, doAs=false, isOpen=true, isDefault=true,
expires in 588529859ms, WM state poolName=null, clusterFraction=0.0,
queryId=null, killReason=null
2018-01-24T15:07:31,746 INFO [HiveServer2-Background-Pool: Thread-1377]
tez.WmEvent: Added WMEvent: EventType: GET EventStartTimestamp: 1516835251746
elapsedTime: 0 wmTezSessionInfo:SessionId: 2be29c62-9f2c-40b7-a5eb-6298baf83a34
Pool: BI Cluster %: 0.0
2018-01-24T15:07:31,746 INFO [Workload management master]
tez.GuaranteedTasksAllocator: Updating 2be29c62-9f2c-40b7-a5eb-6298baf83a34
with 3 guaranteed tasks
{noformat}
> WM: trigger test may fail
> -------------------------
>
> Key: HIVE-18547
> URL: https://issues.apache.org/jira/browse/HIVE-18547
> Project: Hive
> Issue Type: Bug
> Reporter: Sergey Shelukhin
> Priority: Major
>
> https://builds.apache.org/job/PreCommit-HIVE-Build/8818/testReport/org.apache.hive.jdbc/TestTriggersMoveWorkloadManager/testTriggerMoveAndKill/
> Looks like the cluster allocation assignment and WM event creation race,
> probably because WM returns session to the caller ASAP and then makes the
> changes after that.
> {noformat}
> 'Event: GET Pool: BI Cluster %: 80.00' expected in STDERR capture, but not
> found.
> ...
> 2018-01-24T15:07:31,746 INFO [Workload management master]
> tez.WorkloadManager: Processing changes for pool BI: [BI, query parallelism
> 1, fraction of the cluster 0.800000011920929, fraction used by child pools
> 0.0, active sessions 0, initializing sessions 0]
> 2018-01-24T15:07:31,746 INFO [Workload management master]
> tez.WorkloadManager: Starting 1 queries in pool [BI, query parallelism 1,
> fraction of the cluster 0.800000011920929, fraction used by child pools 0.0,
> active sessions 0, initializing sessions 0]
> 2018-01-24T15:07:31,746 INFO [Workload management master]
> tez.WorkloadManager: Received a session from AM pool
> sessionId=2be29c62-9f2c-40b7-a5eb-6298baf83a34, queueName=default,
> user=hiveptest, doAs=false, isOpen=true, isDefault=true, expires in
> 588529859ms, WM state poolName=null, clusterFraction=0.0, queryId=null,
> killReason=null
> 2018-01-24T15:07:31,746 INFO [HiveServer2-Background-Pool: Thread-1377]
> tez.WmEvent: Added WMEvent: EventType: GET EventStartTimestamp: 1516835251746
> elapsedTime: 0 wmTezSessionInfo:SessionId:
> 2be29c62-9f2c-40b7-a5eb-6298baf83a34 Pool: BI Cluster %: 0.0
> 2018-01-24T15:07:31,746 INFO [Workload management master]
> tez.GuaranteedTasksAllocator: Updating 2be29c62-9f2c-40b7-a5eb-6298baf83a34
> with 3 guaranteed tasks
> {noformat}
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
