[11:30]  * benji needs some coffee.
[12:08] <gary_poster> bac benji call in 2
[12:08] <bac> rt
[13:45] <benji> well, I'm now able to change the code for that stored procedure, but I can't get any debugging info out of it
[13:53] <gary_poster> :-/
[13:55] <gary_poster> We have another new test failure (seemingly not a test isolation error this time).  Which is, you know, just so gosh darn dispiriting that I think the company really ought to support me going over to the local IMAX theater to watch the Avengers this afternoon, just to, you know, buck up my spirits.
[13:58] <benji> heh
[14:14] <gary_poster> huh, that failure happened again: bug 994602
[14:15] <_mup_> Bug #994602: lib/lp/services/webapp/tests/cookie-authentication.txt fails rarely/intermittently in tests <paralleltest> <Launchpad itself:Triaged> < https://launchpad.net/bugs/994602 >
[14:15] <gary_poster> can't dupe locally
[14:44] <gary_poster> babysitting/lunch
[14:45] <gary_poster> wgrant says 994602 is his fault
[14:45] <gary_poster> we also have an instance of "the wrong number of tests are reported" to investigate
[14:46] <gary_poster> I definitely got the subunit output this time
[15:00] <benji> interesting, if I isolate just the test that fails, it doesn't; there is some sort of intra-test isolation problem going on
[15:09] <benji> gary_poster: I just submitted my April EC2 expenses (just under $100)
[15:29] <gary_poster> approved benji
[15:29] <benji> thanks
[16:02] <gary_poster> benji, fun with PG 9.1, eh? :-)
[16:03] <benji> :)
[16:35] <gary_poster> benji, here's something interesting (and unrelated yo your current work): take a glance at http://ec2-184-73-44-105.compute-1.amazonaws.com:8010/builders/lucid_lp/builds/0 .  The number of tests run is low.  Click on the worker-2 log to see why.  (I don't know why that happened.)
[16:36] <benji> gary_poster: that's interesting
[16:37] <benji> gary_poster: the "subunit" log has no entries for worker-2
[16:37] <gary_poster> exactle benji
[16:37] <gary_poster> y
[16:37] <gary_poster> the string "worker-2" is not in the subunit log
[16:38] <gary_poster> benji, ah-ha
[16:38] <gary_poster> http://ec2-184-73-44-105.compute-1.amazonaws.com:8010/builders/lucid_lp/builds/0/steps/shell_8/logs/stdio
[16:39] <gary_poster> near the top
[16:39] <gary_poster> everything seems to be going fine, with 8 lxc-start-ephemerals
[16:39] <gary_poster> but then "could not get IP address - aborting."
[16:39] <gary_poster> and "Stopping lxc"
[16:39] <benji> "could not get IP address - aborting.
[16:39] <benji> "
[16:39] <gary_poster> lxcip may help with that
[16:40] <gary_poster> or maybe we just need a bigger timeout
[16:40] <benji> feels race-y
[16:40] <gary_poster> I'm not sure we are racing anything
[16:40] <gary_poster> just things too longer than expected
[16:42] <benji> it's a timed race ;)
[16:42] <gary_poster> benji, look at "less `which lxc-start-ephemeral`" (or choose your voibng poison of course) and search for ""could not get IP address - aborting."
[16:43] <benji> If only I knew what the heck "[ 0 -eq $? -a -n "$IP_ADDRESS" ]" means.
[16:45] <gary_poster> "the last exit code was 0 and we have an ip address"
[16:45] <gary_poster> so the line before that if statement failed
[16:46] <gary_poster> we should retry that
[16:46] <gary_poster> rather than just giving up immediately
[16:46] <gary_poster> ideally we'd have lxcip
[16:46] <gary_poster> since that does everything for us in a nicer way
[16:47] <benji> so we're loosing a race with $LEASES being populated, retrying seems eminently reasonable
[16:47] <gary_poster> yeah
[16:48] <gary_poster> I'll talk to hallyn about it
[17:01]  * benji (long) lunches.
[17:33] <gmb> gary_poster, Can we schedule our annual review call for Monday or Wednesday next week? I have free time in the mornings, and it seems to make sense to get it done next week rather than taking up hacking time with it.
[17:34] <gary_poster> hey gmb.  +1.  Monday would be slightly easier but Wed is fine too.  Choose a time that's not too late, you west coaster you. :-)
[17:35] <gary_poster> gmb, just put it on the Google calendar?
[17:35] <gmb> gary_poster, Yep, I will do. Monday is fine.
[17:36] <gary_poster> cool, thank you
[17:37] <gmb> gary_poster, Done.
[17:38] <gary_poster> accepted, gmb
[17:38] <gmb> Thanks.
[19:43] <gary_poster> benji, out of morbid, look-at-the-crash-on-the-side-of-the-road curiosity, have you managed to get postgres 9.1 working?
[19:44] <benji> gary_poster: I gave up trying to upgrade (a clone of) my lxc lucid container (which in hindsight wasn't a good idea anyway) and am building a new precise container now
[19:45] <gary_poster> benji, a precise container?  postgres 9.1 won't work in lucid?
[19:46] <benji> gary_poster: it might but I exceeded my self-imposed timebox without getting it to work
[19:46] <gary_poster> I see
[19:46] <gary_poster> switching to precise introduces so many other variables though...
[19:46] <gary_poster> and if you get it working there we have to figure out how to get it working in lucid anyway
[19:47] <benji> gary_poster: do you think it won't work out of the box?
[19:47] <gary_poster> benji, what is "it" in that sentence? :-)
[19:47] <benji> gary_poster: Everything!!  :P
[19:48] <gary_poster> lol
[19:48] <benji> (LP)
[19:48] <benji> (LP on precise)
[19:48] <benji> (LP on precise with postgres 9.1)
[19:49] <benji> I was under the (apparently wrong) impression that precise was the best bet for a working LP.  I get the feeling that lucid would be better.
[19:49] <gary_poster> benji, so you are asking if I think LP will work on Precise?  wgrant recently closed a bug for getting LP to run on precise, so I suspect it will.
[19:49] <gary_poster> it has only been working in precise for a week or two at most
[19:49] <gary_poster> and we are still running LP on Lucid
[19:49] <benji> I regret my decision.
[19:49] <gary_poster> :-)
[19:50] <gary_poster> running LP on Lucid in production I mean
[19:51] <benji> I'm goign to kill the precise setuplxc and start a new lucid one in the hope that it will a) work, and b) install postgres 9.1 by default (which I think it will)
[19:51] <gary_poster> benji, since you've already gone down that road it might still be interesting; however, we'll need to get it working in Lucid anyway if this actually fixes anything.  (Or we have to switch production and our whole lxc setup in containers to Precise)
[19:51] <gary_poster> but that sounds like a good plan too
[19:52] <gary_poster> I mean, ugh
[19:52] <gary_poster> I like voice dictation :-P
[19:52] <gary_poster> You *could* keep trying the precise road; might be interesting.  However the Lucid plan does sound better to me.
[19:53] <gary_poster> There we go.
[19:53] <benji> you need a stenographer
[19:53] <gary_poster> I'll hire one posthastee
[19:53] <gary_poster> eeee
[19:54] <gary_poster> I've heard the whole stenographer biz has changed a lot just in the past year or so
[19:54] <benji> been reading the stenographer trade mags again?
[19:54] <gary_poster> computerized voice recognition has changed the stenographer's job to being more like an editor
[19:55] <gary_poster> no, a mom at the elementary school does that part time and was talking about it :-)
[19:59] <benji> gary_poster: well, with this running I have some time if you're wanting to do the yearly review call
[20:00] <gary_poster> benji, sure.  4:05?  That will give me a chance to prepare.  By which I mean, uh, preparing.  Uh, never mind.  $:-5?
[20:00] <gary_poster> 4:05?
[20:00] <benji> heh, sure
[20:00] <gary_poster> cool
[20:06] <gary_poster> benji, https://talkgadget.google.com/hangouts/_/extras/canonical.com/goldenhordeoneonone awaits
[21:20] <bac> have a good weekend gentlemens
[21:54] <gary_poster> you too
[21:54] <gary_poster> bye