Basically, this is a copy of the Buildbot configs running on staging-master:/builds/buildbot/hg-unittest-master (note that this is locally patched from the mozilla2-unittest configs). I've modified the steps slightly to checkout mozconfigs from the new location, not run an HgPoller of its own, etc. The mozconfigs have been made as consistent as the dep/nightly ones as possible. The Builders have been renamed to be consistent with their dep/nightly/leak test counterparts. The soon-to-be-created slaves have been added to the slavelists for dep/nightly/leak test + unittest builds. The commented out win32 builder has been completely removed. Most importantly, unittests will now be run on the same master as the rest of our builds. I think that's everything. It's a big looking patch, but the bulk of unittest_master.cfg is unchanged. Lukas, I'm mainly asking you for review to make sure I haven't accidentally tossed away any of the important local changes. (Though I don't think I tossed any of them away at all).

Comment on attachment 343258 [details] [diff] [review] unittest and builds in the same master Just add more slaves, up to slave18 on linux and win32.

So, let's also add bm-xserve22 to the Mac list

changeset: 422:cb3b8ba6c989

Over to Lukas for slave setup.

Turns out I forgot to import isHgPollerTriggered, and used config instead of nightly_config.

Comment on attachment 343914 [details] [diff] [review] [checked in] follow-up to fix on-change traceback changeset: 424:6fe13c512ae4

Just noticed on moz2-win32-slave04 that I had to agree to the pskill license the first time a build run. This should be done on _all_ of windows slaves before we run them through buildbot.

My bad, I accepted as Admin on all of them...They are now all accepted as cltbld. (In reply to comment #9) > Just noticed on moz2-win32-slave04 that I had to agree to the pskill license > the first time a build run. This should be done on _all_ of windows slaves > before we run them through buildbot.

Comment on attachment 346684 [details] [diff] [review] [checked in] set flunkOnFailure=False for unittest changeset: 492:8c675383696e

Was talking with joduinn and for now, until these are working smoothly, we should keep the steps the same as they are on the unittest master, so checking out mozconfig with cvs instead of hg for now.

Comment on attachment 346761 [details] [diff] [review] check out mozconfig with cvs, not hg As part of merging the unittests with the regular builds we should be pulling the mozconfigs from the same place IMHO. Let's keep the mozconfigs in sync, yes, but pulling them from HG is the right thing to do and it's not going to cause any mysterious failures.

Comment on attachment 349244 [details] [diff] [review] Fix reload problems Checking in steps.py; /cvsroot/mozilla/tools/buildbotcustom/unittest/steps.py,v <-- steps.py new revision: 1.15; previous revision: 1.14 done

So I think it's time to close this, mozilla2-staging buildbot is running fine and the standalone staging for unittest has been turned off.