08:05:19 <anteaya> #startmeeting third-party
08:05:20 <openstack> Meeting started Tue Aug 18 08:05:19 2015 UTC and is due to finish in 60 minutes.  The chair is anteaya. Information about MeetBot at http://wiki.debian.org/MeetBot.
08:05:21 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
08:05:24 <openstack> The meeting name has been set to 'third_party'
08:05:42 <anteaya> hello hello if anyone is here
08:05:49 <lennyb> Hi
08:06:15 <eantyshev> o/
08:06:34 <eantyshev> hello!
08:07:17 <anteaya> hello lennyb and eantyshev
08:07:22 <anteaya> how are you?
08:07:39 <lennyb> thanks, how are you?
08:08:19 <anteaya> good thanks
08:08:30 <anteaya> what shall we discuss today?
08:09:29 <lennyb> if we have nothing to discuss I have a question
08:10:20 <eantyshev> I have nothing particular important
08:10:21 <anteaya> go ahead
08:10:26 <lennyb> I heard there is a way to trigger my CI after it passed some basic gates. where can I read more about it?
08:10:30 <anteaya> eantyshev: okay thannks
08:10:41 <anteaya> I don't know whaat you mean lennyb
08:10:56 <anteaya> can you try to explain what you are talking about?
08:11:03 <anteaya> sorry if I am being slow
08:11:12 <anteaya> can you just try again?
08:11:19 <eantyshev> lennyb, do you mean after Openstack Jenkins completed its check?
08:11:46 <lennyb> I would like to be triggered not by gerrit commit, but by pass of let's say gate-system-config-pep8
08:12:03 <lennyb> eantishev: yes, I guess
08:12:25 <anteaya> I don't know where to read about doing thata
08:12:38 <lennyb> eantishev: after Jenkins completed with success
08:12:41 <anteaya> as that isn't something I would advocate for
08:12:59 <eantyshev> I think I know what you mean, our CI is triggered by comment of 'jenkins' user including 'Builb succeeded' message, will give you a piece of config now
08:13:00 <anteaya> some pepole do configure there systems that way and I can't stop them
08:13:17 <anteaya> but I personally am not a supporter of that direction
08:13:36 <eantyshev> we do that because we have to
08:14:08 <lennyb> anteaya: why? I dont see a reason to load git and our setups if Jenkins failed.
08:14:17 <eantyshev> or there will be too much computation for our CI
08:14:51 <anteaya> right that is the arguement from ci operators
08:15:59 <anteaya> the counter argument is that information inn't available earlier in the patch cycle to be incoporated either from reviews by the ci operators on the patch or by the author or human reviewers
08:16:11 <anteaya> I'm not stopping you
08:16:20 <anteaya> but I don't personally support it
08:16:35 <anteaya> and we don't have documentation around that configuration
08:17:19 <lennyb> anteaya: thanks. I see your point.
08:17:40 <anteaya> okay thanks
08:17:42 <lennyb> eantyshev: thanks about configuration tip
08:17:53 <eantyshev> lennyb: http://paste.openstack.org/show/420156/
08:17:53 <anteaya> that was all I was trying to share
08:18:15 <eantyshev> that's how ours works
08:19:22 <eantyshev> anteaya, could you please explain more, I don't understand
08:19:42 <anteaya> lennyb: can you try to explain my point to eantyshev?
08:19:52 <anteaya> and I can try to fill in any holes
08:21:28 <lennyb> eantyshev: I think the point is to provide as much info and as as quickly as possible to developer.
08:22:05 <anteaya> lennyb: thank you, yes that is the point
08:22:39 <anteaya> also that there is a difference between what infra supports and what operators choose to do
08:23:06 <anteaya> so should an operator choose to do something that is not the process supported by infra
08:23:23 <anteaya> and a patch merges that breakes the operator's ci system
08:23:24 <lennyb> anteaya: how much time usually takes to Jenkins to comment the patch?
08:23:35 <anteaya> that responsibility rests with the ci operator
08:23:51 <anteaya> depends on how many tests are run
08:24:04 <anteaya> pep8 and lint jobs are fast
08:24:19 <anteaya> unit tests take as long as unit tests take
08:24:29 <eantyshev> lennyb: usually 1 hour for nova and cinder which we check, but sometimes longer
08:24:32 <anteaya> and functional and integration tests can take over an hour
08:24:46 <anteaya> depends on the individual test job
08:25:00 <anteaya> we like to support jobs running in about an hour
08:25:10 <anteaya> but there are jobs that are more than that
08:25:21 <anteaya> because that is how long it takes
08:25:51 <anteaya> but we do encourage review of how jobs are built by individual repos
08:26:04 <anteaya> and encourage refactoring to shave off time
08:26:16 <anteaya> and infra works to reduce build time as well
08:27:16 <eantyshev> anteaya: I understand that concerns, recent merge failures of some CIs is a consequence of this optimization
08:27:56 <anteaya> sorry can you expand on your point?
08:29:22 <eantyshev> those CIs which are triggered by Jenkins comments, have higher probability to fail to merge the patch
08:30:33 <anteaya> ah
08:31:29 <anteaya> I see
08:31:42 <anteaya> did we want to discuss this topic further?
08:31:46 <anteaya> or are we done?
08:32:13 <lennyb> I have nothing else
08:32:21 <eantyshev> another thing I want to mention
08:32:52 <eantyshev> I have created https://review.openstack.org/#/c/207094 some time ago
08:33:07 <eantyshev> and I need another +2 on that
08:33:13 <anteaya> #link https://review.openstack.org/#/c/207094
08:34:01 <anteaya> yes you do
08:34:15 <anteaya> I councel patience
08:34:19 <eantyshev> I beleive this should improve the experience of third-party CI operators, as this enables them to check cross-repo dependencies
08:34:37 <anteaya> it is really hard to get all the things reviewed when we want them in
08:34:52 <anteaya> I belive your perspective
08:35:12 <anteaya> hopefully jeblair will have time to look at it soon
08:35:16 <eantyshev> thanks
08:35:22 <anteaya> he did just buy a house today
08:35:42 <anteaya> so he will be very busy offline for the next while doing home new ownership things
08:35:55 <anteaya> but hopefully your patch will be reviewed soon
08:36:02 <anteaya> thanks for bringing it up
08:36:08 <anteaya> anything more today?
08:36:13 <eantyshev> no
08:36:20 <anteaya> okay thank you
08:36:30 <anteaya> so nice to talk to you both every week
08:36:36 <anteaya> :)
08:36:43 <anteaya> enjoy the rest of your day
08:36:48 <anteaya> and see you next week
08:36:52 <anteaya> #endmeeting