launchpad-dev team mailing list archive

[Michael Hudson <michael.hudson@xxxxxxxxxxxxx>]

Francis J. Lacoste wrote:
> On November 2, 2009, Francis J. Lacoste wrote:
>> It got through, but unfortunately, Gary, Michael and Stefan are trying to 
>> unwedge the buildbot at about the same time, which resulted in new failed 
>> builds getting us back in [testfix] mode.
>>
>> So emails are working and once they sort out the buildbot issue, we should
>>  be  back on track.
>>
> 
> We are out of testfix mode and the devel builder is working fine.
> 
> I submitted a manual merge of devel into db-devel. Unfortunately, the db_lp 
> wasn't substantiated following that merge.

Actually, it looks like buildbot never saw this merge; it never even
tried to substantiate the db_lp builder in response to it.... becasue
your merge wasn't a [testfix].  Grr.  I'll send it a testfix build now.

> The jscheck builder got 
> substantiated correctly and is building with the new revision (unfortunately, 
> that's not really helping us.)
> 
> db_lp is still wedged. Fortunately, it's not wedged in testfix mode this time.
> 
> So we should land all r-c patches possible through devel for the time being.
> 
> Michael (H.) could you see if you can debug this with Steve? Please let us 
> know the status of this at the end of your shift.

Steve is on leave today, so the situation at the end of my shift is
going to be the same as it was at the end of yours: completely confused.

Here's what we need to do to find out some more clues:

1) Wait until the db_lp builder isn't building, if necessary.
1) Get a losa ready
2) Force a build on the db_lp builder
3) If the instance is showing as running in elasticfox or equivalent but
not connected in buildbot, get the losa to ssh in to it.
4) Copy the twistd.log files somewhere more accessible.  They'll be in
/srv/buildbot/slaves/launchpad/twistd.log or some such.

Cheers,
mwh