Monday, 2018-06-11

*** zaneb has quit IRC00:06
*** zaneb has joined #openstack-tc00:18
*** kumarmn has joined #openstack-tc00:59
*** kumarmn has quit IRC01:04
*** zaneb has quit IRC01:10
*** zaneb has joined #openstack-tc01:20
*** kumarmn has joined #openstack-tc01:29
*** kumarmn has quit IRC01:35
*** zaneb has quit IRC01:36
*** zaneb has joined #openstack-tc01:41
*** edmondsw has quit IRC01:47
*** kumarmn has joined #openstack-tc02:19
*** kumarmn has quit IRC02:24
*** kumarmn has joined #openstack-tc02:39
*** kumarmn has quit IRC02:45
*** edmondsw has joined #openstack-tc02:54
*** edmondsw has quit IRC02:59
*** kumarmn has joined #openstack-tc03:16
*** kumarmn has quit IRC03:20
*** kumarmn has joined #openstack-tc03:22
*** kumarmn_ has joined #openstack-tc03:26
*** kumarmn has quit IRC03:29
*** kumarmn_ has quit IRC04:21
*** gagehugo has quit IRC04:24
*** dklyle has joined #openstack-tc04:25
*** dklyle has quit IRC04:31
*** gagehugo has joined #openstack-tc04:37
*** kumarmn has joined #openstack-tc04:51
*** kumarmn has quit IRC04:55
*** kumarmn has joined #openstack-tc05:31
*** kumarmn has quit IRC05:36
*** kumarmn has joined #openstack-tc06:04
*** kumarmn has quit IRC06:09
*** jaosorior has joined #openstack-tc07:10
*** kumarmn has joined #openstack-tc07:15
*** kumarmn has quit IRC07:20
*** bauzas has joined #openstack-tc07:39
*** jpich has joined #openstack-tc08:00
*** dtantsur|afk is now known as dtantsur08:40
*** dtantsur is now known as dtantsur|bbl09:49
*** kumarmn has joined #openstack-tc10:38
*** kumarmn has quit IRC10:42
*** jrollen is now known as jroll10:44
*** cdent has joined #openstack-tc11:00
*** kumarmn has joined #openstack-tc11:16
*** jroll has quit IRC11:18
*** jroll has joined #openstack-tc11:18
*** kumarmn has quit IRC11:21
*** kumarmn has joined #openstack-tc11:54
*** kumarmn has quit IRC11:59
*** kumarmn has joined #openstack-tc12:03
*** kumarmn has quit IRC12:08
*** edmondsw has joined #openstack-tc12:12
openstackgerritMerged openstack/governance master: Add openstackclient to Chef OpenStack
*** dtantsur|bbl is now known as dtantsur13:25
smcginnisApparently it's "pad your stats day" today -
cdentah, that13:26
cdentwhile I'm sure this is padding by people, the changes themselves are not negative are they?13:28
smcginniscdent: I hate to see all the gate resources consumed for a patch that changes one character in a comment that is understandable without the change. Especially when it's a whole series of patches doing the same one character change in multiple files instead of just one patch.13:29
smcginnisBut yeah, they aren't necessarily "negative".13:29
cdenttrue, good point. Half of me wishes that we just had _way_ more gate resources and then we wouldn't need to worry13:30
fungicould be trying to get their "one required change" landed to get discount summit passes before feature freeze (assuming we even do those for berlin)13:30
smcginnisI've seen the same behavior happening in other communities, so I do think it is all about the metrics.13:31
*** kumarmn has joined #openstack-tc13:32
*** mriedem has joined #openstack-tc13:44
mnaseri think it would be nice if we figure out a way to make jobs run more efficiently13:58
mnaseras in, only run on the things that need to be covered13:58
openstackgerritMerged openstack/governance master: fix tox python3 overrides
mugsieI think a lot of the irrevant files execptions have been good for that13:58
smcginnisWe've gotten a little better with irrelevant-files, but hard to tell with these that do change relevant files but only code comments.13:59
dtroyerI don't think it is even smart stats-padding, its mechanical, I've gotten a couple of those on two starlingx repos already :)14:01
zanebmnaser: zuul allows that (on a per-file basis though, so you can't e.g. avoid running jobs on changes to comments)14:01
*** ianychoi has quit IRC14:02
mugsieI got one on the openstack letsencrypt plugin as well :)14:03
zanebto some extent reviewers are encouraging this stuff by approving pointless changes like "Replace Chinese punctuation with English punctuation" - which has nothing to do with Chinese, and consists of replacing curly quotes in the docs with straight quotes, both of which are rendered identically as curly quotes in the HTML output14:04
mugsiezaneb: ++14:05
mugsieI don't review them14:06
*** zaneb has quit IRC14:09
*** zaneb has joined #openstack-tc14:13
fungithe trick with trying to avoid running jobs against certain changes is that it's non-trivial to guess what file changes can impact what functionality. inevitably as soon as you try to get fancy about it, breaking bugs start slipping through and wedging the jobs you didn't run14:14
*** annabelleB has joined #openstack-tc14:15
*** needssleep is now known as TheJulia14:15
fungialso, rerunning jobs which you don't expect to be impacted by a given change isn't a total waste, as (especially in the gate pipeline) it still gives us useful telemetry on nondeterministic bugs already present in the software14:15
*** ricolin__ has joined #openstack-tc14:21
*** dklyle has joined #openstack-tc14:40
*** hongbin has joined #openstack-tc14:46
*** dklyle has quit IRC14:47
fungismcginnis: on the barbican topic, what i'm talking about there is things like if a feature requires the user to upload key material, then don't create a key upload method in the calling service's api just expect them to upload the key through barbican's api and tell you which key to use14:59
smcginnisfungi: OK, I thought castellan was meant to help keep us from relying on a specific backend implementation in case we wanted to swap out barbican for $OTHER at some point.15:00
fungibarbican already supports multiple backends itself15:00
smcginnisSo why do we use castellan? Why not just go right to barbican then.15:00
jrollthis is what doug was asking a few weeks ago when this came back up :)15:01
cdentas I recall because castellan allowed avoid barbican, so it all sounds a bit tautological?15:01
fungibecause if there are features which don't require user interaction but do need a key store, then deployments can make use of those features without necessarily having to install barbican15:01
fungithis all goes back to the initial conversations around making barbican a base service and the compromise decided on back then (over a year ago now) was to create a library interface15:02
smcginnisThat seems a little odd to me. They still need to deploy something. And then the user experience is it's fine to use whatever is supported, but oh, you want to use that feature? Then you must deploy this other thing.15:03
fungibut i still think the flip-side of that is, if you need a user-facing api to do things barbican provides, then use barbican for that15:03
smcginnisI goes it is rehashing past conversations, but if barbican supports configuring different backends, and we may require barbican for certain functionality, I guess I would rather just drop castellan and go just with barbican.15:04
fungiif your service only needs access to a key store, then use castellan along with a supported key store (which could be barbican, or could be something else)15:05
fungithe argument was "our deployments need a key store for non-user-facing feature x, and we already have vault installed for other reasons, so we'd rather not have to run an additional service with a user-facing api just for that"15:07
cdentthe complexity of managing/turning off/tuning/etc "user-facing api" is the crux, yes?15:08
fungiin those situations, they have the option of also installing barbican and configuring it to use vault as a backend, when the time comes that they have some new security feature they want to make use of which does have a user-facing component handled by the barbican api15:08
fungiand i think in those situations we do want them deploying barbican rather than adding yet still more duplicate apis in random services and passing around sensitive data unnecessarily15:09
fungicdent: yeah, basically it's the fear of openstack's never-ending service/api sprawl15:11
fungii personally think it would be great if we could just get everybody to deploy barbican and be done with it, but there was significant pushback when that got proposed, and i still think that having a key store available for non-user-facing security features is still a net win so i'd rather not unnecessarily delay that with broader-reaching hopes15:13
dtantsurfolks, do we have SB stories for Pike-era project-wide goals?15:32
fungidtantsur: nope, those were tracked in git, why do you ask?15:33
dtantsurfungi: we have a few fallouts since then. wondering where to track them.15:33
dtantsurusing SB would be handy15:33
fungidtantsur: there's an argument underway currently as to whether it makes sense to expect goal completion past the end of the cycle for that goal, or for some other arbitrary duration, or until the end of time15:34
dtantsurwell, sometimes it's not quite easy. e.g. for WSGI goal: one of our two services did not (and still does not) have a separate API service AT ALL15:35
fungion one hand, saying you completed a pike goal when the code you shipped for pike doesn't have it seems a little misleading15:35
dtantsurmm, fair15:36
fungibut also, at what point do we want people to be able to free up the tracking overhead/busywork for goals completed long after the effort was timeboxed15:37
dtantsurmaybe we should not call it "a Pike goal" on SB?15:38
dtantsurlike, on it's associated with Pike, but on SB it's just "Deploy your API with WSGI"15:38
fungiwe previously accepted tracking updates after a goal's cycle, but never decided how long that should be expected to continue, and as we accumulate more historical goals cycle after cycle expecting projects to keep tracking them all seems burdensome15:39
smcginnisWell, if they all complete them, then that historical burden is gone, right? ;)15:40
fungiyeah, it also came up whether we should put in the effort to "import" tracking for all the historical <=pike goals into sb15:40
fungiwhich goes hand-in-hand with whether there's benefit to having teams continue to update year+ old goals15:41
fungialso do we bother to retroactively note whether retired deliverables met those goals? what about recording that deliverables added after the goal's cycle?15:55
fungithe argument in favor of continuing to update them is that you can look at the goal document/story to determine what deliverables meet that goal, but in practice that doesn't hold up unless we indefinitely add all future deliverables to it too15:57
*** dtruong_ has quit IRC16:05
*** ricolin__ has quit IRC16:27
*** jpich has quit IRC16:41
*** annabelleB has quit IRC17:02
*** annabelleB has joined #openstack-tc17:14
*** dklyle has joined #openstack-tc17:32
*** dtantsur is now known as dtantsur|afk17:37
*** kumarmn has quit IRC18:00
*** kumarmn has joined #openstack-tc18:01
*** kumarmn has quit IRC18:10
*** kumarmn has joined #openstack-tc18:11
*** kumarmn has quit IRC18:15
*** dklyle has quit IRC18:17
*** kumarmn has joined #openstack-tc18:24
*** kumarmn has quit IRC18:29
*** kumarmn has joined #openstack-tc18:30
*** dtruong has joined #openstack-tc18:31
*** kumarmn has quit IRC18:35
*** kumarmn has joined #openstack-tc18:36
*** annabelleB has quit IRC18:41
*** kumarmn has quit IRC18:47
*** kumarmn has joined #openstack-tc18:48
*** dklyle has joined #openstack-tc18:48
*** kumarmn has quit IRC18:52
*** kumarmn has joined #openstack-tc18:54
*** kumarmn has quit IRC18:56
*** kumarmn has joined #openstack-tc18:56
*** annabelleB has joined #openstack-tc19:02
*** annabelleB has quit IRC19:20
*** mriedem1 has joined #openstack-tc19:26
*** mriedem has quit IRC19:28
*** mriedem1 is now known as mriedem19:37
*** annabelleB has joined #openstack-tc19:41
-openstackstatus- NOTICE: Zuul was restarted for a software upgrade; changes uploaded or approved between 19:30 and 19:50 will need to be rechecked19:57
*** annabelleB has quit IRC20:11
*** annabelleB has joined #openstack-tc21:17
*** kumarmn has quit IRC21:55
*** annabelleB has quit IRC22:02
*** cdent has quit IRC22:19
*** mriedem has quit IRC22:21
*** kumarmn has joined #openstack-tc22:23
*** kumarmn has quit IRC22:28
*** EmilienM|PTO is now known as EmilienM22:41
*** edmondsw has quit IRC22:47
*** annabelleB has joined #openstack-tc22:49
*** hongbin has quit IRC23:26
*** mriedem has joined #openstack-tc23:31
*** annabelleB has quit IRC23:42
*** mriedem has quit IRC23:51

Generated by 2.15.3 by Marius Gedminas - find it at!