[LU-4819] tests: Lustre Sanity test_101b hangs for large OSTCOUNTs

• Type: Bug
• Resolution: Fixed
• Priority: Minor
• Fix Version/s: Lustre 2.6.0
• Labels: patch
• Severity: 3
• Assignee: Nathaniel Clark
• Reporter: parinay v kondekar (Inactive)

Description

The Wizard appears to hang during test 101b of the Lustre Sanity test. There have been three instances of this type of hang on this particular system.
Xyratex-bug-id: MRP-1362

# ps aux | grep [r]eads
root     121602 99.9  0.0   4040   440 ?        R    Sep12 939:47 reads -f /tmp/tmpNmB5Fv/f101b -l 18432 -o 18304 -s 104857600 -b 1048576 -a 128 -n 0

In all three instances the test logs appear to stop at test 101b:

name: test_101a
            status: PASS
            duration: 1
            return_code: 0
            error:
        -
            name: test_101b

Logs

== sanity test 101b: check stride-io mode read-ahead =================== 15:04:08 (1379023448)
error: set_param: /proc/{fs,sys}/{lnet,lustre}/obdfilter/*/read_cache_enable: Found no match
error: set_param: /proc/{fs,sys}/{lnet,lustre}/obdfilter/*/writethrough_cache_enable: Found no match