The SeaMonkey unit test boxes should be switched to run on comm-central once it goes live, the config for them will move over to hg.mozilla.org/build/buildbot-configs subdir seamonkey.unittest. This needs an upgrade of buildbot to trunk on all those boxes, so I can't set up the new config in parallel, will need to work on that once comm-central is live.

http://hg.mozilla.org/build/buildbot-configs/index.cgi/file/tip/seamonkey-unittest/ has preliminary configs for those builds, modeled largely after the build machines configs and using buildbotcustom. That config is completely untested though, as I only can do that in production once the new comm-central is live.

The config has been updated to a state that runs, but all the boxes are orange now. Linux crashes in mochichrome tests. Mac fails (at least) 1 TUnit test and aborts, fails 1 reftest, and crashes in mochichrome tests. Windows fails (at least) 6 TUnit tests and aborts, and fails 1 mochichrome test.

Update: Linux and Windows are passing all tests (look at the numbers) and are orange only for leaks, bonus points for fixing those! Mac is orange for two failures: 1) TUnit failure on the test from bug 378991 - FF is running 10.5 and succeeding, we are running 10.4 and failing, is this a Leopard vs. Tiger problem? 2) a reftest fails due to a tiny color glitch, see bug 378217 comment #95 for detailed analysis.

(In reply to comment #3) > Update: > > Linux and Windows are passing all tests (look at the numbers) and are orange > only for leaks, bonus points for fixing those! I filed bug 448125. > Mac is orange for two failures: > 1) TUnit failure on the test from bug 378991 - FF is running 10.5 and > succeeding, we are running 10.4 and failing, is this a Leopard vs. Tiger > problem? Mark filed bug 447999. > 2) a reftest fails due to a tiny color glitch, see bug 378217 comment #95 for > detailed analysis. I filed bug 448121. *** I filed bug 448126 too, on a MacOSX (profile related !?) focus issue.

Setting the bug itself to FIXED, as the machines have been successfully switched to comm-central, we have dependent open bugs on the test failures, which are not about wrong setup of the machines, but actual tests failing.

BTW: There is one (new?) reftest failure on the Mac unit test box, it's REFTEST TEST-UNEXPECTED-FAIL | file:///builds/slave/comm-central-macosx/build/mozilla/layout/reftests/forms/input-text-size-1.html

(In reply to comment #6) > BTW: There is one (new?) reftest failure on the Mac unit test box, it's (Not new:) See bug 449533 and its dependence. NB: Don't add reports to this bug, file (blocking) bugs.

No need to tell me how to use Bugzilla.