[jira] (LIVY-541) Multiple Livy servers submitting to Yarn results in LivyException: Session is finished ... No YARN application is found with tag livy-session-197-uveqmqyj in 300 seconds. Please check your cluster status, it is may be very busy

["Larry McCay (Jira)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/LIVY-541 ]


    Larry McCay deleted comment on LIVY-541:
    ----------------------------------

was (Author: lmccay):
Due to the pending 2.0.0 release this JIRA has been pushed out to 2.1.0 as part of a bulk update. If there is a specific reason to pull this back into the 2.0.0 release and you intend to provide a PR in the next few days please provide justification and reset the Fix Version to 2.0.0.


> Multiple Livy servers submitting to Yarn results in LivyException: Session is finished ... No YARN application is found with tag livy-session-197-uveqmqyj in 300 seconds. Please check your cluster status, it is may be very busy
> -----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
>
>                 Key: LIVY-541
>                 URL: https://issues.apache.org/jira/browse/LIVY-541
>             Project: Livy
>          Issue Type: Bug
>          Components: Server
>    Affects Versions: 0.5.0
>         Environment: Hortonworks HDP 2.6
>            Reporter: Hari Sekhon
>            Priority: Critical
>             Fix For: 0.9.0
>
>          Time Spent: 20m
>  Remaining Estimate: 0h
>
> It appears Livy doesn't differentiate sessions properly in Yarn causing errors when running multiple Livy servers behind a load balancer for HA / performance scaling on the same Hadoop cluster.
> Each livy server uses monotonically incrementing session IDs with a random suffix but it appears that the random suffix isn't passed to Yarn which results in the following errors on the Livy server which is further behind in session numbers because it appears to find the session with the same number has already finished (submitted earlier by a different user on another Livy server as seen in Yarn RM UI):
> {code:java}
> org.apache.zeppelin.livy.LivyException: Session 197 is finished, appId: null, log: [	at org.apache.hadoop.fs.FileSystem$Cache.closeAll(FileSystem.java:2887), at org.apache.hadoop.fs.FileSystem$Cache$ClientFinalizer.run(FileSystem.java:2904), at java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:511), at java.util.concurrent.FutureTask.run(FutureTask.java:266), at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142), at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617), at java.lang.Thread.run(Thread.java:748), 
> YARN Diagnostics: , java.lang.Exception: No YARN application is found with tag livy-session-197-uveqmqyj in 300 seconds. Please check your cluster status, it is may be very busy., org.apache.livy.utils.SparkYarnApp.org$apache$livy$utils$SparkYarnApp$$getAppIdFromTag(SparkYarnApp.scala:182) org.apache.livy.utils.SparkYarnApp$$anonfun$1$$anonfun$4.apply(SparkYarnApp.scala:239) org.apache.livy.utils.SparkYarnApp$$anonfun$1$$anonfun$4.apply(SparkYarnApp.scala:236) scala.Option.getOrElse(Option.scala:120) org.apache.livy.utils.SparkYarnApp$$anonfun$1.apply$mcV$sp(SparkYarnApp.scala:236) org.apache.livy.Utils$$anon$1.run(Utils.scala:94)]
> at org.apache.zeppelin.livy.BaseLivyInterpreter.createSession(BaseLivyInterpreter.java:300)
> at org.apache.zeppelin.livy.BaseLivyInterpreter.initLivySession(BaseLivyInterpreter.java:184)
> at org.apache.zeppelin.livy.LivySharedInterpreter.open(LivySharedInterpreter.java:57)
> at org.apache.zeppelin.interpreter.LazyOpenInterpreter.open(LazyOpenInterpreter.java:69)
> at org.apache.zeppelin.livy.BaseLivyInterpreter.getLivySharedInterpreter(BaseLivyInterpreter.java:165)
> at org.apache.zeppelin.livy.BaseLivyInterpreter.open(BaseLivyInterpreter.java:139)
> at org.apache.zeppelin.interpreter.LazyOpenInterpreter.open(LazyOpenInterpreter.java:69)
> at org.apache.zeppelin.interpreter.remote.RemoteInterpreterServer$InterpretJob.jobRun(RemoteInterpreterServer.java:493)
> at org.apache.zeppelin.scheduler.Job.run(Job.java:175)
> at org.apache.zeppelin.scheduler.FIFOScheduler$1.run(FIFOScheduler.java:139)
> at java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:511)
> at java.util.concurrent.FutureTask.run(FutureTask.java:266)
> at java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.access$201(ScheduledThreadPoolExecutor.java:180)
> at java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.run(ScheduledThreadPoolExecutor.java:293)
> at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
> at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
> at java.lang.Thread.run(Thread.java:748)
> {code}



--
This message was sent by Atlassian Jira
(v8.20.10#820010)