| *** saneax is now known as saneax-_-|AFK | 00:30 | |
| openstackgerrit | Merged openstack-infra/nodepool: Manage providers in ProviderWorker threads https://review.openstack.org/424808 | 00:48 |
|---|---|---|
| jeblair | Shrews: i +3d https://review.openstack.org/427353 with a comment | 00:49 |
| openstackgerrit | Merged openstack-infra/nodepool: Add storeNode ZK API method https://review.openstack.org/427353 | 00:54 |
| Shrews | jeblair: thx. will take care of commenting it | 01:08 |
| *** Shuo has quit IRC | 01:10 | |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul: Run pre and post playbooks https://review.openstack.org/427494 | 01:20 |
| jeblair | mordred: that patch is still going to fail a couple of the unit tests; i have a solution for it but it's going to take a bit. however, it's unrelated to anything in the ansible launcher, so it's safe for you to start using as a base. tests.unit.test_v3.TestAnsible works. | 01:21 |
| mordred | jeblair: cool | 02:56 |
| *** piccobit_ has joined #zuul | 05:42 | |
| piccobit_ | jeblair: hi, good morning! made some progress in making Gearman also triggering Pipeline builds, but currently it doesn't stop triggering an endless loop of builds. any hints what might get wrong? | 05:42 |
| *** saneax-_-|AFK is now known as saneax | 06:05 | |
| *** bhavik1 has joined #zuul | 06:13 | |
| *** abregman has joined #zuul | 06:20 | |
| *** yolanda has quit IRC | 06:23 | |
| *** bhavik1 has quit IRC | 06:36 | |
| *** piccobit has joined #zuul | 06:48 | |
| *** piccobit_ has quit IRC | 06:49 | |
| *** abregman has quit IRC | 06:53 | |
| *** abregman has joined #zuul | 06:55 | |
| openstackgerrit | Tobias Henkel proposed openstack-infra/zuul: Cloner: Better infrastructure failure handling https://review.openstack.org/403559 | 07:09 |
| *** piccobit has quit IRC | 07:44 | |
| *** yolanda has joined #zuul | 08:04 | |
| *** abregman has quit IRC | 08:46 | |
| *** hashar has joined #zuul | 09:30 | |
| *** saneax is now known as saneax-_-|AFK | 09:33 | |
| *** saneax-_-|AFK is now known as saneax | 10:38 | |
| *** piccobit has joined #zuul | 11:21 | |
| *** jamielennox is now known as jamielennox|away | 11:25 | |
| *** hashar is now known as hasharAway | 11:27 | |
| *** threebean has left #zuul | 11:56 | |
| *** hasharAway has quit IRC | 12:32 | |
| *** hashar has joined #zuul | 12:45 | |
| *** mordred has quit IRC | 13:10 | |
| *** adam_g has quit IRC | 13:17 | |
| *** mordred has joined #zuul | 13:18 | |
| *** adam_g has joined #zuul | 13:25 | |
| *** hashar has quit IRC | 13:39 | |
| *** hashar has joined #zuul | 15:00 | |
| *** saneax is now known as saneax-_-|AFK | 15:17 | |
| openstackgerrit | Monty Taylor proposed openstack-infra/nodepool: Remove need for findNetwork https://review.openstack.org/398462 | 15:51 |
| mordred | Shrews: in your RequestHandler patch, I love the comment "poll handlers" followe by the comment "TODO implement polling" :) | 15:52 |
| Shrews | mordred: yeah. i actually *have* it implemented but didn't want to include more than that so as to keep it reviewable | 15:55 |
| Shrews | (well, mostly implemented) | 15:55 |
| mordred | Shrews: that sounds like when I "implement" things | 15:56 |
| Shrews | that review is part of breaking up a much larger change | 15:56 |
| Shrews | ZOMG: http://www.wral.com/plummeting-pork-reserves-threaten-nationwide-bacon-supply/16491597/ | 16:05 |
| Zara | argh I just tabbed past here and now I'm hungry | 16:07 |
| mordred | Zara: it's frustrating in multiple directions isn't it? | 16:08 |
| Zara | ;_; | 16:09 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul: Run pre and post playbooks https://review.openstack.org/427494 | 16:49 |
| jeblair | mordred: okay, that should take care of the remaining failing tests | 16:50 |
| mordred | jeblair: woot | 16:51 |
| *** piccobit has quit IRC | 16:57 | |
| *** yolanda has quit IRC | 17:01 | |
| *** piccobit has joined #zuul | 17:19 | |
| *** bhavik1 has joined #zuul | 17:20 | |
| *** Shuo has joined #zuul | 17:33 | |
| *** bhavik1 has quit IRC | 17:34 | |
| *** nt has quit IRC | 17:46 | |
| *** hashar_ has joined #zuul | 17:58 | |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul: Run pre and post playbooks https://review.openstack.org/427494 | 18:00 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul: Fix test_live_configuration https://review.openstack.org/427838 | 18:00 |
| *** hashar_ has quit IRC | 18:03 | |
| *** hashar has quit IRC | 18:05 | |
| *** harlowja_ has joined #zuul | 18:05 | |
| *** harlowja has quit IRC | 18:06 | |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul: Add SourceContext class https://review.openstack.org/427846 | 18:21 |
| jeblair | mordred: anything happen with that testtools pr? | 18:21 |
| mordred | jeblair: remind me what that was? oh - I made a pr to testtools right? | 18:30 |
| jeblair | yeah, about attaching debug logs only on failure | 18:31 |
| mordred | yah- I need to fix the pr somehow | 18:33 |
| mordred | but haven't had a chance yet | 18:33 |
| jeblair | ok, np. | 18:33 |
| jeblair | it's not as important as the security context thing :) | 18:34 |
| *** yolanda has joined #zuul | 18:39 | |
| *** piccobit has quit IRC | 19:05 | |
| *** harlowja_ has quit IRC | 19:19 | |
| *** hashar has joined #zuul | 19:49 | |
| openstackgerrit | David Shrewsbury proposed openstack-infra/nodepool: Fix documentation nits from earlier reviews. https://review.openstack.org/427884 | 19:51 |
| *** tobiash has quit IRC | 20:14 | |
| *** tobiash has joined #zuul | 20:16 | |
| *** jamielennox|away is now known as jamielennox | 20:30 | |
| *** nt has joined #zuul | 20:35 | |
| *** harlowja has joined #zuul | 20:39 | |
| *** hashar has quit IRC | 21:09 | |
| *** rattboi is now known as ratboi | 21:48 | |
| *** ratboi is now known as rattboi | 21:49 | |
| pabelanger | slowly getting back online from devconf.cz travels. Should be ready to loop back to zuulv3 things tomorrow | 21:59 |
| jhesketh | Morning | 21:59 |
| mordred | look. it's a jhesketh AND a pabelanger | 22:08 |
| mordred | jeblair: with the actually-run-playbooks changes, do we have any concept of 'zuul standard library' of playbooks separate from config repos and project repos? | 22:10 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul: Add SourceContext class https://review.openstack.org/427846 | 22:10 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul: Run pre and post playbooks https://review.openstack.org/427494 | 22:10 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul: Tests: store debug logs on error https://review.openstack.org/427946 | 22:10 |
| jeblair | mordred: not yet -- is it time to figure that out yet? (i also need to finish up the patch to add roles to jobs; that may be related?) | 22:11 |
| mordred | jeblair: I don't _think_ it's time to figure it out necessarily? I was just chatting with dkranz about "examples of zuul jobs" and it struck me that several of the simple examples are quite likely to be examples of consuming content from a standard library | 22:12 |
| jeblair | mordred, clarkb, morgan, Shrews: 427946 may be of interest -- i think it gets close to what i want in my heart of hearts -- full debug logs iff a test fails. | 22:12 |
| mordred | jeblair: oooh | 22:12 |
| jeblair | mordred: ok, i think we're going to need stdlib soon, especially since we want to try to implement "push git repos to node" as a pre-playbook invoking that from stdlib. so maybe let's pencil it in for after roles? | 22:13 |
| mordred | jeblair: wfm | 22:14 |
| mordred | jeblair: one thing about it that might be different is the association between where a job is defined and where a playbook is defined might not make sense if there is a stdlib of playbooks - since a consumer may want to define jobs in their config repo that use those playbooks | 22:15 |
| * mordred waves hands | 22:15 | |
| * mordred waves feet | 22:16 | |
| * mordred jumps around and falls over, skinning his knee | 22:16 | |
| morgan | jeblair: looking | 22:17 |
| morgan | jeblair: oh that is nice. | 22:18 |
| morgan | jeblair: yes good stuff. | 22:18 |
| jeblair | morgan: mordred found the addOnException thing which is key :) | 22:19 |
| morgan | yes, yes it is | 22:19 |
| mordred | jeblair: btw - my patch to the fixtures library has the problem that fixtures do not imply testtools and addOnException is a testtools thing so ASFKNWERFAE RVEAORV | 22:20 |
| morgan | jeblair: but omg yes, logging improvements! | 22:20 |
| jeblair | mordred: some of that might be roles? but also, when i think 'stdlib' i think of something that's internal to zuul. so far in my mind, 'stdlib' means something like "zuul-push-repos-to-host" which is built into zuul. and then "site-local-stuff that gets used a lot" might be roles or a playbook in a common-config repo. maybe there's a third thing, but maybe use cases can slot into one of those two? | 22:20 |
| jeblair | mordred: oh. sigh. maybe this easiest then... it could probably be made into a fixture and librarified... | 22:21 |
| mordred | jeblair: yah - for me stdlib is things like "playbook to run puppet" "playbook to run tox" "playbook to do ./configure ; make ; make distcheck" | 22:21 |
| morgan | mordred: ++ on stdlib things | 22:21 |
| mordred | jeblair: basically, stuff we'd totally put into project-config for openstack but that is actually common stuff that normal humans would want too | 22:21 |
| jeblair | mordred: agreed those make sense for stdlib | 22:21 |
| jeblair | mordred: so if we build that into zuul, we don't have to worry so much about fitting it into the job-playbook system | 22:22 |
| mordred | jeblair: yah. | 22:23 |
| mordred | oh - speaking of (but not really) I _really_ need to make an ansible module which is "please install the packages in this bindep file" | 22:23 |
| jeblair | mordred: my forthcoming patch to make job playbook specification optional fits with that too, i think -- we can say "run: zuul-puppet-playbook-thing" | 22:23 |
| morgan | mordred: oh that would be a NICE ansible thing | 22:23 |
| mordred | morgan: yah - the use case keeps coming up for folks "what's the best way for me to write a playbook that installs packages on redhat and ubuntu in a sane way" | 22:24 |
| mordred | jeblair: cool | 22:24 |
| jeblair | mordred: having said all that, we should be careful we don't abstract ansible out too much -- the *real* power here is that you can write a playbook that uses the puppet role, not that zuul has a playbook that runs the puppet role. :) | 22:26 |
| mordred | jeblair: indeed :) | 22:27 |
| mordred | jeblair: I think the biggest win for 'stdlib' things would be for things users would not find value in having or writing an ansible playbook to do. configure ; make ; make distcheck comes to mind in that category - it's neat that you can ansible it - but nobody doing those things is likely to find that playbook easier | 22:29 |
| mordred | (or "please push these zuul git repos to this computer" :) ) | 22:30 |
| jeblair | mordred: yep, and i still agree. just thinking that at some point, we're going to notice a line in the sand that we have to decide whether or not to cross :) | 22:30 |
| mordred | oh golly yes | 22:30 |
| mordred | I mean, we'll probably discover the line in the sand the normal way - which is me running right past it, then going "oops" | 22:30 |
| morgan | mordred: are you going to setup a volleyball net over that line when you find it? | 22:35 |
| morgan | so others don't run past it, and are inclined to stop and hang out around it :P | 22:36 |
| * jeblair lays tripwire | 22:36 | |
| mordred | morgan: that sounds like a lot of work | 22:46 |
| mordred | morgan: I'm much more likely to just complain and then do some sulking | 22:46 |
| morgan | mordred: hehe | 22:46 |
| mordred | jhesketh: Shrews had a question for you in https://review.openstack.org/#/c/427393 and I'm also curious as to the answer | 22:52 |
| Shrews | mordred, jhesketh: re: the advantage of poll vs callback, poll is already implemented in unsubmitted code, so that's the only advantage i see :) | 23:23 |
| Shrews | plus, the handler can't trigger the callback since it's not a thread itself, so the parent thread really needs to trigger it | 23:25 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul: Tests: store debug logs on error https://review.openstack.org/427946 | 23:37 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul: Add SourceContext class https://review.openstack.org/427846 | 23:37 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul: Run pre and post playbooks https://review.openstack.org/427494 | 23:37 |
| jeblair | Shrews, mordred, jhesketh: yeah, from a flow pov, the handler is a state machine that's driven by the main thread, so whether it's polling or callbacks, action is initiated there. callbacks could be a different approach for handling the accounting (main thread -> state machine -> callback to main provider object to clean up state machine) but i don't think it would substantially change the program flow. | 23:38 |
| SpamapS | jeblair: FYI, https://lists.cncf.io/pipermail/cncf-ci-public/2017-February/000002.html | 23:51 |
| SpamapS | jeblair: I'm pimping Zuul to CNCF ;) | 23:51 |
| jhesketh | mordred, Shrews, jeblair: added a comment to the change... if the code is already implemented then I wouldn't bother doing anything else :-) | 23:55 |
| mordred | jhesketh: cool! I was going to merge the change but didn't want to without resolution (seemed like a relevant/pertinent question :) ) | 23:56 |
| jhesketh | thanks :-) | 23:56 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!