Closed Bug 650865 Opened 14 years ago Closed 14 years ago

add pid check for runtestsremote to cleanup.py run

Categories

(Release Engineering :: General, defect, P4)

ARM
Android
defect

Tracking

(Not tracked)

RESOLVED FIXED

People

(Reporter: bear, Assigned: bear)

References

Details

(Whiteboard: [android][tegra][mobile_unittests])

Attachments

(2 files, 3 obsolete files)

when bug 650828 lands add check for these pids to the cleanup.py step so that it can be noticed *before* the tests start
Depends on: 650828
Whiteboard: [android][tegra][mobile_unittests]
Priority: -- → P4
Blocks: 649215
Blocks: 650535
Assignee: nobody → bear
Attachment #527282 - Flags: review?(aki)
Attachment #527283 - Flags: review?(aki)
Comment on attachment 527282 [details] [diff] [review] add pidfile parameter to remote test calls Have you tried this in staging? By my reading of process/factory.py, basedir is getting set to the slavedir, which is /builds/tegra-###. You're setting the pidfile to /builds/remotereftest.pid ? Or am I reading this wrong?
Comment on attachment 527283 [details] [diff] [review] add pidfile check to cleanup.py step >+pidDir = os.path.join(cwd, '..') I want to doublecheck that this is aligned with the builder (see previous comment). >+for f in ('runtestsremote', 'remotereftest'): >+ pidFile = os.path.join(pidDir, '%s.pid' % f) >+ if os.path.exists(pidFile): >+ print "pidfile from prior test run found, trying to kill" >+ stopProcess(pidFile, f) >+ if not os.path.exists(pidFile): >+ setFlag(errorFile, "Remote Device Error: process from previous test run present [%s]" % f) >+ sys.exit(2) if *not* os.path.exists(pidFile) it's an error? I think this is a bug.
(In reply to comment #4) > Comment on attachment 527283 [details] [diff] [review] > add pidfile check to cleanup.py step > > >+pidDir = os.path.join(cwd, '..') > > I want to doublecheck that this is aligned with the builder (see previous > comment). yea, I was unsure of this and when I checked the tinderbox output I saw that the basedir property was shown as /builds/tegra-###/test I'll be running this in staging to find out > > >+for f in ('runtestsremote', 'remotereftest'): > >+ pidFile = os.path.join(pidDir, '%s.pid' % f) > >+ if os.path.exists(pidFile): > >+ print "pidfile from prior test run found, trying to kill" > >+ stopProcess(pidFile, f) > >+ if not os.path.exists(pidFile): > >+ setFlag(errorFile, "Remote Device Error: process from previous test run present [%s]" % f) > >+ sys.exit(2) > > if *not* os.path.exists(pidFile) it's an error? > I think this is a bug. the "if not" part is because after running stopProcess, the pid file is still around, then we were unable to .... doh! good catch
changed path to pidfile to move it to /builds/tegra-###
Attachment #527282 - Attachment is obsolete: true
Attachment #527282 - Flags: review?(aki)
Attachment #527381 - Flags: review?(aki)
adjusted cleanup.py to pidfile path change and also sneaking in a bugfix to stopProcess()
Attachment #527283 - Attachment is obsolete: true
Attachment #527283 - Flags: review?(aki)
Attachment #527382 - Flags: review?(aki)
Comment on attachment 527381 [details] [diff] [review] add pidfile parameter to remote test calls Per IRC, basedir is being set to /builds/tegra-###/test, so the pidfiles will go in /builds/tegra-###. r=me
Attachment #527381 - Flags: review?(aki) → review+
Attachment #527382 - Attachment is obsolete: true
Attachment #527382 - Flags: review?(aki)
Attachment #527387 - Flags: review?(aki)
Attachment #527387 - Flags: review?(aki) → review+
Comment on attachment 527381 [details] [diff] [review] add pidfile parameter to remote test calls committed changeset 1495:a8404b0bcbe4
Attachment #527381 - Flags: checked-in+
Comment on attachment 527387 [details] [diff] [review] add pidfile check to cleanup.py step committed changeset 1371:4c03d0266f86 this change is waiting for the buildbotcustom change to land before it will be active
Attachment #527387 - Flags: checked-in+
buildbotcustom change landed in reconfig done 21 Apr 2011
Status: NEW → RESOLVED
Closed: 14 years ago
Resolution: --- → FIXED
Product: mozilla.org → Release Engineering
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: