Failing simple map-reduce jobs due to task failure

[Saeed Adel Mehraban <s....@gmail.com>]

I have a Hadoop 2.2.0 installation on 3 VMs, one as master and 2 as slaves.
When I try to run simple jobs like provided wordcount sample, if I try to
run the job on 1 or a few files, it probably will succeed or not (somehow
50-50 chance of failure) but with more files, I get failure most of the
time.
Job's syslog are attached and I will provide you whatever files you need to
know about, just help me understand what is the problem.
Regards.