You should report progress in a period less then configured (in you case
600sec).
Add code like below to you reducer and call ping in you reducer where you
process tuples.
final TaskAttemptContext context = <init in costructor>;
long lastTime = System.currentTimeMillis();
public void ping() {
final long currtime = System.currentTimeMillis();
if (currtime - lastTime > 10000) {
context.progress();
lastTime = currtime;
}
}
2010/5/11 Corbin Hoenes <[email protected]>
> Not sure I am clean on how I can debug stuff on a cluster. I currently
> have a long running reducer that attempts to run 4 times before finally
> giving up
>
> I get 4 of these: Task attempt_201005101345_0052_r_000012_0 failed to
> report status for 601 seconds. Killing!
>
> before it gives up...on the last try I noticed this in the log:
> ERROR: org.apache.hadoop.hdfs.DFSClient - Exception closing file
> /tmp/temp1925356068/tmp1003826561/_temporary/_attempt_201005101345_0052_r_000012_4/abs/tmp/temp1925356068/tmp-197182389/part-00012
> : org.apache.hadoop.ipc.RemoteException: java.io.IOException: Could not
> complete write to file
> /tmp/temp1925356068/tmp1003826561/_temporary/_attempt_201005101345_0052_r_000012_4/abs/tmp/temp1925356068/tmp-197182389/part-00012
> by DFSClient_attempt_201005101345_0052_r_000012_4
> at
> org.apache.hadoop.hdfs.server.namenode.NameNode.complete(NameNode.java:497)
> at sun.reflect.GeneratedMethodAccessor23.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$Server.call(RPC.java:508)
> at org.apache.hadoop.ipc.Server$Handler$1.run(Server.java:966)
> at org.apache.hadoop.ipc.Server$Handler$1.run(Server.java:962)
> at java.security.AccessController.doPrivileged(Native Method)
> at javax.security.auth.Subject.doAs(Subject.java:396)
> at org.apache.hadoop.ipc.Server$Handler.run(Server.java:960)
> How do I turn on log4j's DEBUG statements? Hoping those will help me
> pinpoint what is going on here--maybe it's the cluster or maybe the script.
>
>
>
