440536 - qm-xserve01 is aborting during tests

Status: RESOLVED FIXED

(Infrastructure & Operations :: RelOps: General, task, P2)

Description

[Daniel Holbert [:dholbert]]

The last 5 runs of qm-xserve01 (a Firefox 3.0 tinderbox) have all been very short orange blips.

The tinderbox only shows "128/7" for the TUnit tests (as opposed to "544/0" on the last green cycle) and 0/0/0 or FAIL for all the other test suites.

The short log for these cycles shows a number of aborts occurring, e.g.
====
../../../_tests/xpcshell-simple/test_necko/test/test_cern_meta.js: /builds/slave/trunk_osx/mozilla/tools/test-harness/xpcshell-simple/test_all.sh: line 129: 25590 Abort trap              NATIVE_TOPSRCDIR="$native_topsrcdir" TOPSRCDIR="$topsrcdir" $xpcshell -s $headfiles -f $t $tailfiles 2>$t.log 1>&2
FAIL
====

Here's a log for one of the bad cycles:
http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1213901400.1213901638.29126.gz#err0


Last green cycle was:
http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1213805940.1213807458.26564.gz

It looks like there were no checkins between the most recent green cycle and the start of the orange blips, so I'm guessing this is a server issue.  (the box probably needs a clobber / reboot / something.)

Comment 1

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

This machine was a bit troubled up till early on the 18th, with nagios reporting that buildbot wasn't running and yet test cycles were still showing up on the tinderbox. Restarting buildbot from a fresh terminal obviously wasn't enough to unconfuse it, as it's now behaving as if buildbot wasn't started via VNC (which it was).

Anyway, I just rebooted the sucker and restarted buildbot. I'll check on it in the morning, but feel free to grab this if need be.

Comment 2

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

You know, it PASSed make check and is passing reftests so reopen if problems start up again.