Closed Bug 523186 Opened 15 years ago Closed 15 years ago

Request for a test image access

Categories

(Release Engineering :: General, defect, P3)

x86
Windows XP
defect

Tracking

(Not tracked)

RESOLVED FIXED

People

(Reporter: jimm, Assigned: nthomas)

References

Details

I'm currently trying to track down leaks caused by a patch for the 1.9.1 branch (bug 470487) that only show up on try server. Curious if it's possible to get access to a test image similar to try's "WINNT 5.2 try hg unit test" for testing purposes to try and track this down.
I'm moving this over to releng.
Assignee: server-ops → nobody
Component: Server Operations → Release Engineering
QA Contact: mrz → release
is this reproducible on all slaves or a subset?
(In reply to comment #2)
> is this reproducible on all slaves or a subset?

WINNT 5.2 try hg unit test AFAICT. It's windows only.
WINNT 5.2 try hg unit test is actually a type of job that can be completed by any one of a number of machines in our pool.  You should see something like "COMPUTERNAME=TRY-W32-SLAVE18" in the environment printout preceding the different build steps.  Have you done this more than once on try?
(In reply to comment #4)
> WINNT 5.2 try hg unit test is actually a type of job that can be completed by
> any one of a number of machines in our pool.  You should see something like
> "COMPUTERNAME=TRY-W32-SLAVE18" in the environment printout preceding the
> different build steps. 

COMPUTERNAME=TRY-W32-SLAVE04

> Have you done this more than once on try?

Yes, multiple times, the leaks always show up.
(In reply to comment #5)
> (In reply to comment #4)
> > WINNT 5.2 try hg unit test is actually a type of job that can be completed by
> > any one of a number of machines in our pool.  You should see something like
> > "COMPUTERNAME=TRY-W32-SLAVE18" in the environment printout preceding the
> > different build steps. 
> 
> COMPUTERNAME=TRY-W32-SLAVE04
> 
> > Have you done this more than once on try?
> 
> Yes, multiple times, the leaks always show up.

Ran it today to be sure - 

COMPUTERNAME=TRY-W32-SLAVE06

same leaks.
We'll lend you try-w32-slave06 for a short while. Please file a bug against mozilla.org::Server Operations if you don't already have access to the MPT VPN.
Assignee: nobody → nthomas
Priority: -- → P2
Renamed buildbot.tac, created a snapshot, scrubbed, and emailed auth to jimm. Rebooting the box may reset the auth and lock you out, but you shouldn't need to do that anyway.

Leaving this open to restore when Jim is done (please let us know!).  Note to self - restoring is revert snapshot, rename buildbot.tac.bak.
Priority: P2 → P3
(In reply to comment #8)
> Leaving this open to restore when Jim is done (please let us know!).  Note to
> self - restoring is revert snapshot, rename buildbot.tac.bak.

And re-enable notifications on the buildbot check in nagios.
Ok, all finished! You can take it back.
Thanks Jim. Snapshot reverted & deleted, and connected back to the try master.

What did we learn from this ? Was full access required to solve the problem ? Would an objdir or objdir/dist have been sufficient ?
Status: NEW → RESOLVED
Closed: 15 years ago
Resolution: --- → FIXED
(In reply to comment #11)
> Thanks Jim. Snapshot reverted & deleted, and connected back to the try master.
> 
> What did we learn from this ? Was full access required to solve the problem ?
> Would an objdir or objdir/dist have been sufficient ?

I used it to track down a bunch of tests that caused a leak. The assumption was it was something specific to our build machines. It ended up related to something quirky in the JVM we have installed. I suppose if I had a few other folks try and reproduce, we probably would have found it on some other machine. Having access to the try slave though made the process simple since I was the guy working on the original patch.
Product: mozilla.org → Release Engineering
You need to log in before you can comment on or make changes to this bug.