552337 - bm-xserve21 may be unwell

Status: RESOLVED WORKSFORME

(Release Engineering :: General, defect, P2)

Description

[Nick Thomas [:nthomas] (UTC+12)]

Three failures doing test jobs

* OS X 10.5.2 addonsmgr debug test mochitests-5/5 build 15, at Sun Mar 14 01:49:13 2010, unpacking tests
  [ lots of unpacking ]
  reftest/tests/layout/reftests/backgrounds/background-size-body-contain.html
  reftest/tests/layout/reftests/backgrounds/background-size-body-cover-no-repeat.html
bzip2: I/O or other error, bailing out.  Possible reason follows.
  process killed by signal 30

* OS X 10.5.2 tracemonkey opt test mochitests-5/5 build 334, at Sun Mar 14 12:19:03 2010, unpacking the build
  [rsyncing files from mount point]
  sent 89570011 bytes  received 4820 bytes  9428929.58 bytes/sec
total size is 89542651  speedup is 1.00
  2010-03-14 12:23:14.202 diskimages-helper[364:1553] -processKernelRequest: will sleep received
  2010-03-14 12:23:14.203 diskimages-helper[364:1553] -processKernelRequest: idle received
  "disk3" unmounted.
  "disk3" unmounted.
  diskimages-helper: DI_kextDriveGetRequest returned 0x00000025 (37) ((os/kern) object terminated).
  "disk3" ejected.
  "disk3" ejected.
  mnt
  process killed by signal 30

* OS X 10.5.2 addonsmgr opt test mochitest-other build 21, at Sun Mar 14 14:45:09 2010, unpacking the build as above

Reimage if this continues ?

Comment 1

[Nick Thomas [:nthomas] (UTC+12)]

Meant to note that these failures are the minority of the test jobs this machine is doing, most run green. Also that signal 30 is SIGUSR1, which we've seen before but not isolated the cause of.

Comment 2

[[e-mail] (:bhearsum)]

I'll try to look into this this week

Comment 3

[[e-mail] (:bhearsum)]

I don't think I'll be able to look at this, after all.

Comment 4

[Chris Cooper [:coop] (he/him)]

Since Mar 14th, bm-xserve21 has had 20 build failures/warnings that might have been caused by machine issues. Of the 8 builds that still have logs I can get at, I've found:

* a bunch of UNEXPECTED fails due to test timeouts in addons-related tests on the addonsmgr branch (not surprising);
* a bunch of legit compile failure on broken code in e10s and tracemonkey;
* 1 bus error: http://production-master02.build.mozilla.org:8010/builders/OS%20X%2010.5.2%20tracemonkey%20build/builds/1187/steps/check/logs/stdio

I'm not seeing a pattern here. Let's keep the bus error in mind and revisit this if we see anything else.