I don't believe the local job tracker can handle launching multiple jobs in
different threads. I am running into the same problems you are, where one job
steps on the temp directory of another.
On Jul 1, 2011, at 5:42 AM, Yaozhen Pan wrote:
Hi,
I am not sure if this question (as title) has
I looked into this a bit more. It seems the LocalJobTracker by default uses a
single local map reduce directory (named localRunner) for all jobs. Thus two
simultaneous jobs will step on each other.
You can work around this however. Set the mapped.local.dir property on each
of the job's
Sorry one more correction, the class is named LocalJobRunner (not
LocalJobTracker).
On Aug 19, 2011, at 4:28 PM, Bryan Keller wrote:
I looked into this a bit more. It seems the LocalJobTracker by default uses a
single local map reduce directory (named localRunner) for all jobs. Thus
two
Correction, the property is mapred.local.dir. (Darn Lion autocorrect).
On Aug 19, 2011, at 4:28 PM, Bryan Keller wrote:
I looked into this a bit more. It seems the LocalJobTracker by default uses a
single local map reduce directory (named localRunner) for all jobs. Thus
two simultaneous
Hi,
I am not sure if this question (as title) has been asked before, but I
didn't find an answer by googling.
I'd like to explain the scenario of my problem:
My program launches several threads in the same time, while each thread will
submit a hadoop job and wait for the job to complete.
The