Job failures because of 'Counters Exceeded limit'

[amitjaiswal]

I am using scoobi_2.10-0.8.0-cdh3-SNAPSHOT and seeing failures in job because of too many counters.

org.apache.hadoop.ipc.RemoteException: java.io.IOException: Counters Exceeded limit: 120 at org.apache.hadoop.mapred.JobTracker.getJobCounters(JobTracker.java:4034) at sun.reflect.GeneratedMethodAccessor48.invoke(Unknown Source) at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:25) at java.lang.reflect.Method.invoke(Method.java:597) at org.apache.hadoop.ipc.RPC$Server.call(RPC.java:587) at org.apache.hadoop.ipc.Server$Handler$1.run(Server.java:1444) at org.apache.hadoop.ipc.Server$Handler$1.run(Server.java:1440) at java.security.AccessController.doPrivileged(Native Method) at javax.security.auth.Subject.doAs(Subject.java:396) at org.apache.hadoop.security.UserGroupInformation.doAs(UserGroupInformation.java:1232) at org.apache.hadoop.ipc.Server$Handler.run(Server.java:1438)

This is happening either during or at the end of the job execution. On some debugging, I found that the scoobi jobs are creating a counter unique for each map task like:

scoobi.counter.mapper.values mapper-0 0

Is it possible that when all the counters are aggregated, then the total number of counters exceeds the default limit leading to job failure?

Can somebody please look into this.

Thanks. Amit

[etorreborre]

I'm looking into this. Thanks for reporting the issue.

[etorreborre]

BTW, do you know that you can increase the number of counters per job by using the mapreduce.job.counters.limit property (the default is 120)?

[etorreborre]

Please wait a bit for the new SNAPSHOT to be published.