Occasional Errno 10 with RedirectedExecutable

[Allyn Dimock]

Using unmodified qmtest 2.3, python 2.2 I am testing some programs as

class RunProgramTest(qm.test.test.Test):
   ...
   def Run(self, context, result):
   ...
    e = qm.executable.RedirectedExecutable(self.timeout)
    status = e.Run(arguments, dir=test_home_dir, path=path)

where self.timeout > 0

I occasionally get
   File "/home/bits8/streamit/regtest/20051203.220013.Sat/streams/QMTest/streamit.py", 
line 310, in Run
     status = e.Run(arguments, dir=test_home_dir, path=path)
   File "/usr/uns/encap/qm-2.3/lib/python2.2/site-packages/qm/executable.py", line 548, in Run
     path)
   File "/usr/uns/encap/qm-2.3/lib/python2.2/site-packages/qm/executable.py", line 268, in Run
     status = os.waitpid(child, 0)[1]

This happens with a subset of the tested programs, some of the time.

The programs that it happens to are all ones that spawn multiple threads 
and communicate through sockets (they are designed to run on up to 1 CPU 
per thread across a cluster, but are built for this test to all run on a 
single CPU).  I don't know if this is relevant:  I looked at /proc during 
one failing run and didn't see socket used getting anywhere close to the 
maximum, nor memory usage or anything else that I could think of.

The actual run command on a 686 system with 2 hyperthreaded CPUs is:

qmtest run -j4

The OS environment is RHEL3 and the kernel is 2.4.21-32.0.1.ELsmp

If anyone else has reported a similar problem and / or you have an idea as 
to what is going on, please let me know.  I am sending this unoficially, 
because I seem to have misplaced the URL and my password for your 
bugzilla.

-- Allyn Dimock