A few random thoughts:
Do other map reduce jobs see this problem when running over the same
data set?
Are you using a custom Slicer? This is invoked to decide how many
maps to create, so if there's a problem there, it might look like this.
Is your input data set large? I've run pig scripts that load 12T or
so of data, and it takes several minutes to spawn the ~95K maps that
result from that.
Alan.
On Mar 23, 2009, at 7:18 PM, Vadim Zaliva wrote:
I am experiencing strange problem. I am trying to start a PIG task in
mapreduce mode and it just hangs there.
Job tracker does not show any tasks running and last message on
console is:
2009-03-23 19:10:28,636 [main] INFO
org
.apache
.pig.backend.hadoop.executionengine.mapReduceLayer.MapReduceLauncher
- 0% complete
strace on java process shows that it hangs in:
futex(0xc449e0, FUTEX_WAIT, 15992, NULL
Any hints on how I can debug this?
Sincerely,
Vadim
