Closed
Bug 389336
Opened 17 years ago
Closed 17 years ago
jprof tinderbox broken
Categories
(Core :: General, defect, P2)
Tracking
()
RESOLVED
FIXED
People
(Reporter: ted, Assigned: coop)
References
()
Details
I went to look at rhelmer's tinderbox that uploads jprof data, but all the output files contained was: mmap: Success It's possible jprof is broken, or something in the way the tinderbox is using it.
Comment 1•17 years ago
|
||
Looks like it's burning right now: http://tinderbox.mozilla.org/MozillaTest/ I'll see if the nightlies have been working or not; it should not be pushing data unless it has a green build and the tests don't crash.
Comment 2•17 years ago
|
||
I see this in the logs; checking it out further. JPROF_RTC setup: open("/dev/rtc", O_RDONLY): Device or resource busy jprof: Error initializing RTC, NOT profiling (Gecko:7300): GnomeUI-WARNING **: While connecting to session manager: Authentication Rejected, reason : None of the authentication protocols specified are supported and host-based authentication failed. Could not start jprof-profiling since JPROF_FLAGS was not set.
Status: NEW → ASSIGNED
Updated•17 years ago
|
Assignee: nobody → rhelmer
Status: ASSIGNED → NEW
Comment 3•17 years ago
|
||
Oh :/ There were a ton of hung builds and crash reporter processes, so /dev/rtc was in use. Ted how can I turn off and/or auto-submit crashes in breakpad? :)
Status: NEW → ASSIGNED
Reporter | ||
Comment 4•17 years ago
|
||
MOZ_CRASHREPORTER_DISABLE=1 if you don't need to run the unit test, MOZ_CRASHREPORTER_NO_REPORT=1 if you do. It's all right here in this handy pamphlet: http://developer.mozilla.org/en/docs/Environment_variables_affecting_crash_reporting
Reporter | ||
Updated•17 years ago
|
Summary: jprof broken? → jprof tinderbox broken
Updated•17 years ago
|
Priority: -- → P2
Whiteboard: ETA: August 2nd
Updated•17 years ago
|
Assignee: rhelmer → nobody
Status: ASSIGNED → NEW
Whiteboard: ETA: August 2nd
Assignee | ||
Comment 5•17 years ago
|
||
I'll see if I can ressurect this machine.
Assignee: nobody → ccooper
Assignee | ||
Updated•17 years ago
|
Status: NEW → ASSIGNED
Assignee | ||
Comment 6•17 years ago
|
||
I've set MOZ_CRASHREPORTER_NO_REPORT=1 in cltbld's ENV (via .bash_profile) and clobbered the build.
Assignee | ||
Comment 7•17 years ago
|
||
I honestly have no idea how the X server was setup previously such that cltbld might actually be able to attach to it. It was running as root in runlevel 5 with respawn on, and I certainly couldn't get tinderbox to attach to it using the standard bag of tricks. I've now switched it to runlevel 3 and have it running metacity as cltbld. This needs to be started by hand on reboot, but there is an .xinitrc file present so startx should just work. This is the same setup as on bl-bldlnx03 so at least two linux machines on this network will be using the same linux X strategy now.
Assignee | ||
Updated•17 years ago
|
Status: ASSIGNED → RESOLVED
Closed: 17 years ago
Resolution: --- → FIXED
Comment 8•17 years ago
|
||
I noticed this still isn't working; Couple more things here: 1) /dev/rtc read-only * set rtc to 666 in /etc/udev/permissions.d/50-udev.permissions 2) gnome-segv was using /dev/rtc (probably due to gnome catching the crash since breakpad is turned off) I fixed this in the middle of a run, but I see some data now so I am hopeful that it'll actually work the next run. We need a way to keep an eye on this.
Status: RESOLVED → REOPENED
Resolution: FIXED → ---
Updated•17 years ago
|
Status: REOPENED → ASSIGNED
Comment 9•17 years ago
|
||
We've got data now, looks good again: http://stage.mozilla.org/pub/mozilla.org/firefox/tinderbox-builds/bl-bldlnx02-trunk/
Status: ASSIGNED → RESOLVED
Closed: 17 years ago → 17 years ago
Resolution: --- → FIXED
You need to log in
before you can comment on or make changes to this bug.
Description
•