Bug 875628

Created attachment 643325 [details]
SOSREPORT

V7 memory test on RHEL 6.3 x86_64 failed on 4 TB RAM

Description of problem:
  V7 memory test on RHEL 6.3 x86_64 failed on 4 TB RAM

The first part of the memory test ( 160 x 25 GB ) worked fine:
  <output name="Limits" description="Get test parameters based on hardware">
  System Memory: 4136318 MB
  Free Memory: 4096661 MB
  Swap Memory: 2047 MB
  <summary>PASS</summary>

But the second part ( Single process ) returned “FAIL”:
  <output name="Single-process">
  Starting Threaded Memory Test
  running for more than free memory at 4097685 MB for 60 sec.
  done.
  <summary>FAIL</summary>


Version-Release number of selected component (if applicable):
   Kernel:    2.6.32-220.el6 x86_64
   V7:          v7-1.5-32.el6.noarch


How reproducible:

Steps to Reproduce:
1.install rhel6.3 x86_64
2.install rhel v7 certkit 1.5-32
     ( step 1 and 2 handled by using the kickstart description under
        https://fedorahosted.org/v7/wiki/HwCertHowToKs6_64
3.run v7 memory test by "v7 run --test=memory"
  
Actual results:
   v7 memory test retuned “FAIL”
Expected results:
   v7 memory test should return “PASS”

Additional info:

It seems that the function “threaded_memtest” in “memory.py line 131” returned “fail”.
The function is called with memory = 4097685 MB   <=> 3,90785 TB .

memory.py:
128:            # otherwise
129:            runtime = 60 # sec.
130:            print "running for more than free memory at %u MB for %u sec." % (memory, runtime)
131:            result = self.runCommand("./threaded_memtest -qpv -m%um -t%u" % (memory, runtime))
132:            print "done."
133:            if result != 0: return False

Strace output and SOSREPORT will be attached soon to this bugzilla.

This problem occurred during the 4 TB update certification of our PRIMERGY RX900 S2.
( see also Case #00730679 
  [RHEL 6 x86_64] PRIMERGY RX900 S2 :  4 TB RAM update cert/remote tests