445578 - windows unit test boxes need maintenance after an xpcshell test hangs

Status: RESOLVED FIXED

(Release Engineering :: General, defect)

Description

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

The Windows unit test boxes need manual maintenance if somebody checks in code that causes one of the xpcshell unit tests to hang.  This happens, I think, because the xpcshell process never dies, which means that files are held open, and the tree can't be deleted:

rm -f -rf _tests
rm: cannot unlink `_tests/xpcshell-simple/test_places/autocomplete/head_000.js': Permission denied
rm: cannot unlink `_tests/xpcshell-simple/test_places/autocomplete/head_autocomplete.js': Permission denied
rm: cannot unlink `_tests/xpcshell-simple/test_places/autocomplete/test_417798.js': Permission denied
rm: cannot remove directory `_tests/xpcshell-simple/test_places/autocomplete': Directory not empty
rm: cannot remove directory `_tests/xpcshell-simple/test_places': Directory not empty
make[1]: Leaving directory `/c/builds/slave/win2k3_03/build/objdir'
rm: cannot remove directory `_tests/xpcshell-simple': Directory not empty
rm: cannot remove directory `_tests': Directory not empty
make[1]: *** [default] Error 1
make: *** [build] Error 2


The automation should be fixed so manual intervention (as in bug 445571) is not needed.

Comment 1

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Happened again (bug 446143) due to bug 445893 comment 6.

Comment 2

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

Attachment: Kill xpcshell at start of run

Something like this ?

Comment 3

[Lukas Blakk [:lsblakk] use ?needinfo]

I'm just waiting on some windows vms for moz2 staging so I can run this patch, shouldn't be too much longer.

Comment 4

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

The production machines are hitting this just about every day. Can I do anything to help ?

Comment 5

[Serge Gautherie (:sgautherie)]

(In reply to comment #4)
> The production machines are hitting this just about every day. Can I do
> anything to help ?

Maybe bug 420216 comment 21 !?

Comment 6

[Lukas Blakk [:lsblakk] use ?needinfo]

Comment on attachment 330366 [details] [diff] [review]
Kill xpcshell at start of run

I've used this on staging, will need to get it into all the master.cfg files (total of 4) for unittest

Comment 7

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

Bah, this still doesn't kill the files off all the xpcshell processes. I suspect bug 453260.

Comment 8

[Lukas Blakk [:lsblakk] use ?needinfo]

Not sure if this is still an issue, since bug 420216 is fixed now.  Either way, taking my name off assignment since I'm not actively working on it.  Anyone who wants it can take it.

Comment 9

[John O'Duinn [:joduinn] (please use "needinfo?" flag)]

(In reply to comment #7)
> Bah, this still doesn't kill the files off all the xpcshell processes. I
> suspect bug 453260.

(In reply to comment #8)
> Not sure if this is still an issue, since bug 420216 is fixed now.

hmmm... are we still seeing this happen? 

I'll leave this open for now, but move to Future while we wait-and-see a little longer. If we dont hit this again, lets close this out as done, ok?

Comment 10

[John O'Duinn [:joduinn] (please use "needinfo?" flag)]

(In reply to comment #9)
> (In reply to comment #7)
> (In reply to comment #8)
> > Not sure if this is still an issue, since bug 420216 is fixed now.
> hmmm... are we still seeing this happen? 
> 
> I'll leave this open for now, but move to Future while we wait-and-see a little
> longer. If we dont hit this again, lets close this out as done, ok?

Can we now close this bug?

Comment 11

[[e-mail] (:bhearsum)]

(In reply to comment #10)
> (In reply to comment #9)
> > (In reply to comment #7)
> > (In reply to comment #8)
> > > Not sure if this is still an issue, since bug 420216 is fixed now.
> > hmmm... are we still seeing this happen? 
> > 
> > I'll leave this open for now, but move to Future while we wait-and-see a little
> > longer. If we dont hit this again, lets close this out as done, ok?
> 
> Can we now close this bug?

I'm assuming yes.

Comment 12

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

Moving closed Future bugs into Release Engineering in preparation for removing the Future component.