Hi , all

I wrote a pig script and it works fine. After I add more computation to the existing pig script, the hadoop job count exceed 70, and I get this error from job tracker:
2012-08-08 00:35:22,648 INFO org.apache.hadoop.ipc.Server: IPC Server handler 5 on 8012, call submitJob(job_201208071030_3456, hdfs://DC-hadoop01:9000/home/hadoop/hdfs/hdfstmp/mapred/staging/hadoop/.staging/job_201208071030_3456, org.apache.hadoop.security.Credentials@4ff3b079) from 10.20.100.34:19728: error: java.io.IOException: java.lang.NullPointerException
java.io.IOException: java.lang.NullPointerException
        at org.apache.hadoop.mapred.JobTracker.submitJob(JobTracker.java:3644)
        at org.apache.hadoop.mapred.JobTracker.submitJob(JobTracker.java:3611)
        at sun.reflect.GeneratedMethodAccessor19.invoke(Unknown Source)
        at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
        at java.lang.reflect.Method.invoke(Method.java:601)
        at org.apache.hadoop.ipc.RPC$Server.call(RPC.java:578)
        at org.apache.hadoop.ipc.Server$Handler$1.run(Server.java:1393)
        at org.apache.hadoop.ipc.Server$Handler$1.run(Server.java:1389)
        at java.security.AccessController.doPrivileged(Native Method)
        at javax.security.auth.Subject.doAs(Subject.java:415)
        at org.apache.hadoop.security.UserGroupInformation.doAs(UserGroupInformation.java:1136)
        at org.apache.hadoop.ipc.Server$Handler.run(Server.java:1387)
Caused by: java.lang.NullPointerException
        at org.apache.hadoop.hdfs.DFSClient$DFSOutputStream.closeThreads(DFSClient.java:4026)
        at org.apache.hadoop.hdfs.DFSClient$DFSOutputStream.closeInternal(DFSClient.java:4069)
        at org.apache.hadoop.hdfs.DFSClient$DFSOutputStream.close(DFSClient.java:4004)
        at org.apache.hadoop.hdfs.DFSClient$LeaseChecker.close(DFSClient.java:1514)
        at org.apache.hadoop.hdfs.DFSClient.close(DFSClient.java:383)
        at org.apache.hadoop.hdfs.DistributedFileSystem.close(DistributedFileSystem.java:339)
        at org.apache.hadoop.fs.FileSystem$Cache.closeAll(FileSystem.java:1499)
        at org.apache.hadoop.fs.FileSystem.closeAllForUGI(FileSystem.java:289)
        at org.apache.hadoop.mapred.JobInProgress.<init>(JobInProgress.java:491)
        at org.apache.hadoop.mapred.JobTracker.submitJob(JobTracker.java:3642)

It seems like the DFSOutputStream the job tracker is trying to closing is already closed, so the streamer field is already null.

But I can not turn off the fs cache, since I'm afraid the leaked DFSOutputstream may cause other problems.
Is there any other solution?

Thanks.
Haitao Yao
[EMAIL PROTECTED]
weibo: @haitao_yao
Skype:  haitao.yao.final