| *** jesusaur has quit IRC | 00:12 | |
| *** jesusaur has joined #zuul | 00:40 | |
| openstackgerrit | Adam Gandelman proposed openstack-infra/zuul feature/zuulv3: Re-enable test_json_status https://review.openstack.org/429146 | 05:08 |
|---|---|---|
| openstackgerrit | Merged openstack-infra/zuul feature/zuulv3: Move and re-enable test_build_configuration_branch tests https://review.openstack.org/409376 | 05:26 |
| *** Cibo_ has quit IRC | 06:09 | |
| *** Cibo_ has joined #zuul | 06:31 | |
| *** Cibo_ has quit IRC | 06:35 | |
| *** yolanda has joined #zuul | 06:43 | |
| *** saneax-_-|AFK is now known as saneax | 07:21 | |
| *** hashar has joined #zuul | 09:31 | |
| *** yolanda has quit IRC | 09:56 | |
| *** yolanda has joined #zuul | 09:57 | |
| *** yolanda has quit IRC | 10:26 | |
| *** yolanda_ has joined #zuul | 10:29 | |
| openstackgerrit | Oleksandr Karpenko proposed openstack-infra/zuul master: introduce http authentication for gerrit base_url https://review.openstack.org/429647 | 11:02 |
| *** Cibo_ has joined #zuul | 11:08 | |
| *** yolanda_ has quit IRC | 12:18 | |
| *** yolanda_ has joined #zuul | 12:23 | |
| *** Cibo_ has quit IRC | 12:26 | |
| *** yolanda_ has quit IRC | 12:27 | |
| *** yolanda_ has joined #zuul | 12:28 | |
| *** saneax is now known as saneax-_-|AFK | 12:34 | |
| *** Cibo_ has joined #zuul | 12:39 | |
| *** Cibo_ has quit IRC | 12:44 | |
| *** Cibo_ has joined #zuul | 12:44 | |
| *** Cibo_ has quit IRC | 12:49 | |
| *** Cibo_ has joined #zuul | 12:59 | |
| openstackgerrit | David Shrewsbury proposed openstack-infra/nodepool feature/zuulv3: Assign node set to node requests https://review.openstack.org/428428 | 13:09 |
| Shrews | ^^^ removed the requirement to wait on the node launches during shutdown | 13:11 |
| openstackgerrit | Merged openstack-infra/zuul feature/zuulv3: Avoid multiple build cancelations https://review.openstack.org/428477 | 13:34 |
| openstackgerrit | Merged openstack-infra/zuul feature/zuulv3: Fix nodeset return on skipped jobs https://review.openstack.org/425865 | 13:35 |
| openstackgerrit | Merged openstack-infra/zuul feature/zuulv3: Use playbooks defined in repos https://review.openstack.org/423550 | 13:36 |
| mordred | jeblair: ok - your entire stack is now +A'd - if it all actually passes the gate in a single pass without gerrit getting confused on merges I'm going to be impressed :) | 13:38 |
| Shrews | eek. many -2's from jenkings on those | 13:38 |
| mordred | ok. good that I don't need to be impressed | 13:39 |
| Shrews | woah. that failed test traceback is crazy | 13:39 |
| mordred | yah | 13:40 |
| Shrews | s/traceback/output/ | 13:40 |
| *** pbrobinson has left #zuul | 13:42 | |
| *** saneax-_-|AFK is now known as saneax | 14:10 | |
| *** yolanda_ has quit IRC | 14:21 | |
| *** yolanda_ has joined #zuul | 14:21 | |
| openstackgerrit | Merged openstack-infra/zuul feature/zuulv3: Coalesce zuul and nodepool images in integration https://review.openstack.org/428858 | 14:42 |
| openstackgerrit | Merged openstack-infra/zuul feature/zuulv3: Re-enable test_file_jobs https://review.openstack.org/429115 | 14:42 |
| *** bhavik1 has joined #zuul | 14:58 | |
| *** bhavik1 has quit IRC | 15:07 | |
| *** saneax is now known as saneax-_-|AFK | 15:23 | |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul feature/zuulv3: [WIP] Add reporter for Federated Message Bus (fedmsg) https://review.openstack.org/426861 | 15:28 |
| *** nt has quit IRC | 15:35 | |
| *** herlo has joined #zuul | 15:36 | |
| *** herlo has joined #zuul | 15:36 | |
| *** nt has joined #zuul | 15:36 | |
| *** nt has quit IRC | 15:39 | |
| *** Cibo_ has quit IRC | 15:39 | |
| *** nt has joined #zuul | 15:40 | |
| *** nt has quit IRC | 15:43 | |
| *** nt has joined #zuul | 15:43 | |
| *** nt has quit IRC | 15:44 | |
| *** nt has joined #zuul | 15:44 | |
| *** Cibo_ has joined #zuul | 15:54 | |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul feature/zuulv3: [WIP] Add reporter for Federated Message Bus (fedmsg) https://review.openstack.org/426861 | 15:57 |
| jeblair | mordred, Shrews: i'm just going to recheck any of the ones that fail before the debug log test attachment, since i can't actually debug them. | 16:01 |
| openstackgerrit | Merged openstack-infra/nodepool feature/zuulv3: Replace NodeRequestWorker with NodeRequestHandler https://review.openstack.org/427393 | 16:06 |
| openstackgerrit | Merged openstack-infra/nodepool feature/zuulv3: Fix documentation nits from earlier reviews. https://review.openstack.org/427884 | 16:07 |
| Shrews | jeblair: not sure why i still have WIP in 428428, but it isn't. updating commit msg now | 16:12 |
| Shrews | oh, nm. was on the wrong PS | 16:12 |
| Shrews | gerrit is just silly in showing "related changes" | 16:14 |
| openstackgerrit | Merged openstack-infra/nodepool feature/zuulv3: Fix storeNode() for updates https://review.openstack.org/428425 | 16:20 |
| openstackgerrit | Merged openstack-infra/nodepool feature/zuulv3: Fix race in node request state changes https://review.openstack.org/428741 | 16:20 |
| openstackgerrit | Merged openstack-infra/nodepool feature/zuulv3: Update Node and NodeRequest models https://review.openstack.org/428736 | 16:20 |
| Shrews | w00t | 16:21 |
| openstackgerrit | Merged openstack-infra/zuul feature/zuulv3: Put Zuul vars in an ansible vars file https://review.openstack.org/424367 | 16:42 |
| pabelanger | yay | 16:43 |
| pabelanger | merge all the things | 16:43 |
| openstackgerrit | Merged openstack-infra/zuul feature/zuulv3: Add playbooks for all test jobs https://review.openstack.org/425810 | 16:45 |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul feature/zuulv3: [WIP] Add reporter for Federated Message Bus (fedmsg) https://review.openstack.org/426861 | 16:50 |
| jeblair | i've added an item to today's zuul meeting agenda about preparing for the PTG -- hopefully folks who are planning to show up for that can attend today's meeting and we can talk about what we should do to be ready for that | 16:54 |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul feature/zuulv3: [WIP] Add reporter for Federated Message Bus (fedmsg) https://review.openstack.org/426861 | 17:11 |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul feature/zuulv3: [WIP] Add reporter for Federated Message Bus (fedmsg) https://review.openstack.org/426861 | 17:17 |
| *** hashar is now known as RainbowPickles | 17:18 | |
| *** harlowja has joined #zuul | 17:20 | |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul feature/zuulv3: [WIP] Add reporter for Federated Message Bus (fedmsg) https://review.openstack.org/426861 | 17:25 |
| jeblair | Shrews: this is a thought to just file away for later: i wonder if it would be more efficient to keep track of all of the nodes in-memory in the launcher, and update the in-memory state from zk based on watches in order to get a first-order approximation of the nodes which are ready and could be assigned to requests. we would still have to try to lock them and double check their states before doing so, but that might reduce the number of ... | 17:26 |
| jeblair | ... full-node zk queries we have to do. | 17:26 |
| *** RainbowPickles is now known as hashar | 17:26 | |
| jeblair | Shrews: (that would *definitely* be a case of premature optimization now -- i really thing we should continue on the course you are plotting for now) | 17:26 |
| clarkb | for infra's zuul instance we are talking ~1k node requests per hour. Or ~17 per minute. Seems like zk should be able to handle that many full tree traversals trivially ? | 17:28 |
| jeblair | clarkb: i agree -- another reason not to worry about it now :) | 17:29 |
| jeblair | we'll get the algorithm down first, and then later, we can replace the implementation of "get ready nodes" if we desire. | 17:30 |
| Shrews | i agree with all the thoughts above | 17:30 |
| jeblair | w00t | 17:30 |
| * Shrews currently struggling with the whole label-target stuff around node launching | 17:31 | |
| Shrews | so optimization is not at the forefront of my mind | 17:31 |
| Shrews | :) | 17:31 |
| jeblair | ugh, sorry | 17:31 |
| jeblair | Shrews: oh, ignore target | 17:31 |
| Shrews | but... hostnames | 17:32 |
| Shrews | ?? | 17:32 |
| jeblair | target impacts hostnames? | 17:32 |
| Shrews | http://docs.openstack.org/infra/nodepool/configuration.html#targets | 17:32 |
| Shrews | yes | 17:32 |
| jeblair | yuck | 17:32 |
| jeblair | that should probably be moved to 'label'... | 17:33 |
| jeblair | clarkb, pabelanger: ^ | 17:33 |
| Shrews | i would be very onboard with that | 17:33 |
| mordred | yah. it seems like hostname template should be assocaited with label | 17:34 |
| clarkb | or provider/image | 17:34 |
| clarkb | since thats where things get a name | 17:34 |
| pabelanger | provider seems reasonable to me | 17:34 |
| mordred | yah - well: {label.name}-{provider.name}-{node_id} is hte current default | 17:34 |
| mordred | so whereever it makes sense | 17:35 |
| jeblair | i guess it would work in any of those places, even though it can refer to any of them, since all those values are known when its created... | 17:35 |
| jeblair | 1 sec | 17:35 |
| * jeblair looks up mailing list post on nodepool config structure | 17:36 | |
| jeblair | okay, how about provider for now. that's probably going to be the least amount of repitition for most users. we might also add it to label later. :) | 17:38 |
| pabelanger | ++ | 17:39 |
| jeblair | clarkb, mordred, Shrews: ^? | 17:40 |
| mordred | ++ | 17:40 |
| clarkb | wfm | 17:40 |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul master: Add support for sqlalchemy reporter https://review.openstack.org/223333 | 17:43 |
| pabelanger | jeblair: mordred: ^ just added the tools/test-setup.sh script, needed for database things now. We also need to update JJB for zuul, doing that now | 17:43 |
| Shrews | jeblair: wfm 2 | 17:43 |
| pabelanger | for database support | 17:43 |
| jeblair | pabelanger: thanks | 17:44 |
| pabelanger | actually, JJB might not need updating | 17:44 |
| pabelanger | think we are okay | 17:44 |
| Shrews | jeblair: could you work up a sample that I can work from, just so I know I'm on the same page with you | 17:44 |
| Shrews | ? | 17:44 |
| rcarrillocruz | clarkb: heya, mind approving https://review.openstack.org/#/c/403732/ ? | 17:45 |
| clarkb | rcarrillocruz: ya I plan on reviewing those today as I catch back up again | 17:46 |
| rcarrillocruz | Thx | 17:47 |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul master: Add support for sqlalchemy reporter https://review.openstack.org/223333 | 17:47 |
| jeblair | Shrews: https://etherpad.openstack.org/p/544c45kn5L | 17:47 |
| jeblair | okay, something's going fairly consistently wrong in the tests, so i'm going to reorder my stack to get the debugging stuff earlier. | 17:49 |
| *** harlowja has quit IRC | 17:49 | |
| Shrews | jeblair: thx, and yeah, node_id == node.id seems sane to me | 17:49 |
| jeblair | yeah, there was probably some reason for that which doesn't apply anymore :) | 17:49 |
| jeblair | like maybe the hostname was created before the node record object at some point | 17:50 |
| jeblair | nah, that doesn't make sense either. i have no idea. :) | 17:50 |
| mordred | jeblair: I'm gonna go with hysterical raisins | 17:50 |
| Shrews | jeblair: it should be fine as long as we don't have really long-lived nodes | 17:51 |
| Shrews | since node.id will eventually rollover | 17:51 |
| jeblair | Shrews: even then, if the node still exists, it shouldn't actually collide, right? | 17:52 |
| Shrews | well, nodepool certainly won't care | 17:52 |
| jeblair | hrm. as i think about what we discovered about the zk sequence node protocol, i'm not so sure... | 17:52 |
| jeblair | at some point, we should poke at that, since we will get really big node numbers. but let's back burner that one too. :) | 17:53 |
| Shrews | *nod* that rollover thing has definitely been a (minor) itch in my mind | 17:54 |
| jeblair | sorry: | 17:54 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Add some gearman related debugging https://review.openstack.org/428530 | 17:54 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Set job class attributes in __init__ https://review.openstack.org/428859 | 17:54 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Tests: store debug logs on error https://review.openstack.org/427946 | 17:54 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Add SourceContext class https://review.openstack.org/427846 | 17:54 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Report exceptions in launcher https://review.openstack.org/428499 | 17:54 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Implement job aborting https://review.openstack.org/426306 | 17:54 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Move job running into own class https://review.openstack.org/385964 | 17:54 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Add infrastructure for pre/post playbooks https://review.openstack.org/425989 | 17:54 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Use ZUUL_TEST_ROOT in launcher https://review.openstack.org/428391 | 17:54 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Run pre and post playbooks https://review.openstack.org/427494 | 17:54 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Handle multiple playbook extensions https://review.openstack.org/425450 | 17:54 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Add inheritance debugging https://review.openstack.org/429065 | 17:54 |
| pabelanger | jeblair: mordred: mind +2 on https://review.openstack.org/#/c/223333/ (sqlalchemy reporter) again? Just a rebase and include database setup script. | 17:55 |
| *** hashar has quit IRC | 17:55 | |
| jeblair | pabelanger: why does that script use sudo? | 17:58 |
| pabelanger | jeblair: I would have to confirm with AJaeger, it was copied from nodepool repo | 18:00 |
| jeblair | mordred: oh, is it because of some new mysql thing that requires a root mysql user connection to be root? | 18:00 |
| pabelanger | jeblair: let me see if it is needed | 18:00 |
| openstackgerrit | Paul Belanger proposed openstack-infra/nodepool master: [WIP] Remove sudo access from test-setup.sh https://review.openstack.org/429818 | 18:03 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Add action plugins to restrict untrusted execution https://review.openstack.org/428798 | 18:03 |
| jeblair | oh, i see, there was a collision with a new test-re-enablement which needs playbooks now | 18:09 |
| jeblair | will fix | 18:09 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Add some gearman related debugging https://review.openstack.org/428530 | 18:11 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Set job class attributes in __init__ https://review.openstack.org/428859 | 18:11 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Tests: store debug logs on error https://review.openstack.org/427946 | 18:11 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Add SourceContext class https://review.openstack.org/427846 | 18:11 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Report exceptions in launcher https://review.openstack.org/428499 | 18:11 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Implement job aborting https://review.openstack.org/426306 | 18:11 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Move job running into own class https://review.openstack.org/385964 | 18:11 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Add infrastructure for pre/post playbooks https://review.openstack.org/425989 | 18:11 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Use ZUUL_TEST_ROOT in launcher https://review.openstack.org/428391 | 18:11 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Run pre and post playbooks https://review.openstack.org/427494 | 18:11 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Handle multiple playbook extensions https://review.openstack.org/425450 | 18:11 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Add inheritance debugging https://review.openstack.org/429065 | 18:11 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Add playbooks needed by merge tests https://review.openstack.org/429819 | 18:11 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Add action plugins to restrict untrusted execution https://review.openstack.org/428798 | 18:11 |
| jeblair | mordred, pabelanger: can you +3 429819? | 18:13 |
| jeblair | i reapproved all the others | 18:14 |
| pabelanger | looking | 18:14 |
| pabelanger | +2 | 18:15 |
| jeblair | i'm going to go ahead and approve it since it's trivial and blocking so much. | 18:24 |
| openstackgerrit | Merged openstack-infra/zuul feature/zuulv3: Add playbooks needed by merge tests https://review.openstack.org/429819 | 18:29 |
| *** harlowja has joined #zuul | 18:44 | |
| clarkb | rcarrillocruz: ok reviewed the first one | 18:45 |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul feature/zuulv3: Add reporter for Federated Message Bus (fedmsg) https://review.openstack.org/426861 | 18:50 |
| pabelanger | jeblair: mordred: SpamapS: ^ okay, fedmsg reporter, based on the sql reporter work that jhesketh did. Ready for reviews now. | 18:51 |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul feature/zuulv3: Add reporter for Federated Message Bus (fedmsg) https://review.openstack.org/426861 | 18:54 |
| openstackgerrit | Merged openstack-infra/zuul feature/zuulv3: Tests: store debug logs on error https://review.openstack.org/427946 | 19:04 |
| openstackgerrit | Merged openstack-infra/zuul feature/zuulv3: Handle multiple playbook extensions https://review.openstack.org/425450 | 19:07 |
| openstackgerrit | Merged openstack-infra/zuul feature/zuulv3: Move job running into own class https://review.openstack.org/385964 | 19:08 |
| openstackgerrit | Merged openstack-infra/zuul feature/zuulv3: Add infrastructure for pre/post playbooks https://review.openstack.org/425989 | 19:08 |
| SpamapS | pabelanger: neat | 19:21 |
| SpamapS | pabelanger: if we're adding features to v3 branch tho.. can haz github? ;-) | 19:21 |
| jeblair | i'd really rather us not focus on drivers until we achieve the basics | 19:22 |
| jeblair | pabelanger: it might be nice to WIP that until then | 19:22 |
| pabelanger | jeblair: sure, np | 19:23 |
| SpamapS | me too :) | 19:25 |
| * SpamapS got distracted yak shaving ttrun to see if he could get it to pass along Ctrl-C | 19:26 | |
| jeblair | SpamapS: i often end up using ctrl-\ when things are going poorly to avoid orphans | 19:26 |
| SpamapS | jeblair: no joke, I've never heard of ctrl-\ | 19:28 |
| *** hashar has joined #zuul | 19:29 | |
| jeblair | SpamapS: i think i only learned of it recently myself... i don't recall how. i think it's entirely possible it was a (very surprising) typo -- ctrl-backspace is a think in emacs. | 19:32 |
| rcarrillocruz | clarkb: shell expansion in a var won't work, but we we could do is do a 'getent passwd {{ CI_USER }}' , parse the home folder and move from there | 19:35 |
| rcarrillocruz | iirc, getent belongs to libc package, so should be mostly by default in distros | 19:35 |
| rcarrillocruz | s/we we/what we | 19:35 |
| SpamapS | rcarrillocruz: color me interested... | 19:39 |
| * SpamapS reads | 19:39 | |
| clarkb | rcarrillocruz: I think anything that doesn't assume /home/ holds all the homedirs beacuse at least for root its common not too (though that particular code is a nop if running as root) | 19:40 |
| rcarrillocruz | if we're good with getent, i will push | 19:40 |
| rcarrillocruz | SpamapS: https://review.openstack.org/#/c/403732/13/playbooks/roles/copy_mirror_config/tasks/main.yaml , for context | 19:40 |
| SpamapS | yeah | 19:40 |
| rcarrillocruz | or we can also cat /etc/passwd , grep , cut | 19:40 |
| clarkb | eval echo ~$USER is another potential user | 19:41 |
| rcarrillocruz | i lean towards getent | 19:41 |
| clarkb | s/user/option/ | 19:41 |
| mordred | it would also be not terribly hard to add a jinja filter to ansible that does os.expanduser ... fwiw | 19:41 |
| SpamapS | so this is one thing where I feel like ansible gets shoved into root too quickly because it's too hard to "become" selectively | 19:41 |
| clarkb | SpamapS: I don't think this is a become issue | 19:41 |
| mordred | like, getent for now - but maybe also add an expanduser jinja filter upstream | 19:41 |
| clarkb | its a where is source file issue | 19:41 |
| SpamapS | clarkb: well I'm arguing that it should be easier to just run blocks as "the intended user" and not some unknown admin | 19:42 |
| clarkb | source file has been homedir of user running devstack-gate so we need to preserve that behavior | 19:42 |
| rcarrillocruz | there's expanduser filter, but i think what we wnat here is to know the home of a given user, not the logged on user | 19:42 |
| SpamapS | Like most of the time you don't need root privs | 19:42 |
| clarkb | rcarrillocruz: correct | 19:42 |
| rcarrillocruz | what'd be nice is to have something like 'user_facts', that effectively returns a dict with stuff from /etc/passwd etc | 19:43 |
| mordred | right. but expanduser('~CI_USER') should expand to /home/CI_USER | 19:43 |
| SpamapS | In this particular case you do though. :-/ | 19:43 |
| mordred | because that's what os.expanduser in python does | 19:43 |
| mordred | >>> os.path.expanduser('~mordred') | 19:43 |
| mordred | '/home/mordred' | 19:43 |
| clarkb | SpamapS: this is running as not root | 19:44 |
| openstackgerrit | David Shrewsbury proposed openstack-infra/nodepool feature/zuulv3: Assign node set to node requests https://review.openstack.org/428428 | 19:44 |
| openstackgerrit | David Shrewsbury proposed openstack-infra/nodepool feature/zuulv3: Replace provider template-hostname with hostname https://review.openstack.org/429841 | 19:44 |
| SpamapS | mordred: isn't the question more what happens with os.path.expanduer('~') when you've become root? | 19:44 |
| mordred | nope | 19:44 |
| SpamapS | clarkb: it's writing to /root though? | 19:44 |
| clarkb | SpamapS: we are looking for the source file to copy into roots home dir so that python source installs work properly | 19:44 |
| rcarrillocruz | and that also works if /etc/passwd contains as home for mordred something like /opt/home/mordred ? | 19:44 |
| Shrews | ah crap. didn't mean to update 428428 | 19:44 |
| rcarrillocruz | not sure if expanduser actually reads /etc/passwd ? | 19:44 |
| clarkb | SpamapS: so its not a become issue its a preserve old behavior and copy source file from source location properly issue | 19:44 |
| mordred | rcarrillocruz: yes | 19:45 |
| clarkb | (old code did ~/.pydistutilscfg | 19:45 |
| mordred | rcarrillocruz: I'm pretty sure os.path.expanduser uses getent on the backend | 19:45 |
| rcarrillocruz | ah, then let me try with the filter if it works that way | 19:45 |
| SpamapS | clarkb: old code also selectively sudo'd which I don't see this block doing. | 19:45 |
| SpamapS | which is why I'm confused that it wouldn't be a become issue. | 19:46 |
| clarkb | SpamapS: right but thats not the issue from a functionality position | 19:46 |
| clarkb | we need to preserve the behavior not necessarily the pid for everyting | 19:46 |
| clarkb | er uid | 19:46 |
| SpamapS | how so? Before, I was sudo'ing as $USER so ~ expanded to $USER ... now I'm not sudoing , so I'm root, not $USER, so ~ won't work. | 19:46 |
| clarkb | ~ wont work regardless in ansible | 19:47 |
| clarkb | because you aren't in a shell aiui | 19:47 |
| SpamapS | I think mordred was arguing for emulating it with expanduser() | 19:47 |
| mordred | you are in a shell if you're using 'shell' - you are not if you are using 'command' | 19:47 |
| clarkb | mordred: or the file moduel which I think is in use here | 19:47 |
| mordred | in this case, it's a stat call, so you're right, there is no shell to glob | 19:47 |
| clarkb | anyways my point is lets not get caught up on become when the issue is path related | 19:48 |
| clarkb | we need to preserve behavior | 19:48 |
| clarkb | not necessarily just run the same shell in ansible | 19:48 |
| SpamapS | we're stat'ing and 'command'ing | 19:48 |
| mordred | if you just change from command to shell, then you can just put the ~ into the commands and let the shell do it | 19:48 |
| SpamapS | mordred: except it's running as root, not the user. | 19:48 |
| mordred | but that still doesn't take care of the stat | 19:48 |
| clarkb | SpamapS: so you ~{{CI_USER}} ? | 19:49 |
| SpamapS | Yeah, that. | 19:49 |
| SpamapS | and the stat can use {{ os.path.expanduser() }} or whatever is equivalent in jinja2 land? | 19:49 |
| mordred | yup. exactly | 19:50 |
| SpamapS | or you can set just set_fact | 19:50 |
| SpamapS | and reference the fact directly, which is a bit more DRY | 19:50 |
| SpamapS | mmmm 4 coders all with their own stirring spoons, what could possibly go wrong? | 19:51 |
| * mordred opens a large crate of spoons and starts throwing them at people | 19:51 | |
| SpamapS | ftr I concur with clarkb that the important outcome is preservation of previous behavior, which is useful and intentional. | 19:51 |
| SpamapS | mordred: no, the spoons are not for throwing. | 19:52 |
| pabelanger | there is no spoon | 19:52 |
| clarkb | SpamapS: I bring it up | 19:52 |
| clarkb | er | 19:52 |
| clarkb | I bring it up ebcause its been tricky at times to preserve that behaviopr | 19:52 |
| mordred | SpamapS: this spoon should be green | 19:52 |
| clarkb | reviewing with that specifically in mind has been helpful | 19:52 |
| clarkb | and I know that third party CIs are really good at catching all of teh corners on this stuff so trying to avoid needing to revert for them | 19:54 |
| openstackgerrit | Paul Belanger proposed openstack-infra/nodepool master: [WIP] Remove sudo access from test-setup.sh https://review.openstack.org/429818 | 20:02 |
| openstackgerrit | David Shrewsbury proposed openstack-infra/nodepool feature/zuulv3: Set node state to BUILDING after it is locked. https://review.openstack.org/429848 | 20:17 |
| pabelanger | clarkb: mordred: care to +3 412124? already have 2 +2 | 20:19 |
| mordred | pabelanger: +A | 20:20 |
| Shrews | jeblair: oh, turns out we cannot get rid of template-hostname in the 'provider' section. that, confusingly, is used for the external image name for images uploaded by the builder | 20:22 |
| Shrews | now why that is used for image names is beyond me | 20:23 |
| Shrews | perhaps we should rename it to template-image-name? | 20:23 |
| openstackgerrit | Merged openstack-infra/nodepool master: Set diskimage formats for building state https://review.openstack.org/412124 | 20:24 |
| jeblair | Shrews: 'template' in that case refers to the template host used for snapshots. i guess we can retcon that to mean something else, or call it 'image-name-format' or something | 20:25 |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul feature/zuulv3: Re-enable test_time_database test https://review.openstack.org/429850 | 20:26 |
| pabelanger | easy test reenable for zuulv3^ | 20:26 |
| Shrews | jeblair: image-name-format WFM | 20:27 |
| openstackgerrit | Adam Gandelman proposed openstack-infra/zuul feature/zuulv3: Re-enable test_mutex https://review.openstack.org/429122 | 20:36 |
| pabelanger | adam_g: left a comment on 429146 | 20:36 |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul feature/zuulv3: Re-enable test_client_enqueue_ref test https://review.openstack.org/393887 | 20:39 |
| openstackgerrit | David Shrewsbury proposed openstack-infra/nodepool feature/zuulv3: Add 'hostname' to provider config section https://review.openstack.org/429841 | 20:40 |
| pabelanger | okay, cleaned up my 2 zuul patches for test re-enablement | 20:42 |
| openstackgerrit | David Shrewsbury proposed openstack-infra/nodepool feature/zuulv3: Set node state to BUILDING after it is locked. https://review.openstack.org/429848 | 20:42 |
| adam_g | pabelanger: oh, good question | 20:42 |
| jeblair | Shrews: left some questions/comments on 428428 ps 10. | 20:48 |
| Shrews | jeblair: awesome. thx | 20:49 |
| jeblair | Shrews: and a bikeshed on 429841 | 20:49 |
| Shrews | jeblair: i love bikeshedding | 20:50 |
| jeblair | then you're gonna love it | 20:50 |
| adam_g | is include'ing a python-file no longer a thing in v3? | 20:52 |
| jeblair | adam_g: it is not | 20:52 |
| jeblair | (we're trying to handle all those cases with real configuration) | 20:53 |
| jeblair | Shrews: and one thought on 429848 as well | 20:54 |
| adam_g | jeblair: is there another way to accomplish adding build tags? im looking at https://git.openstack.org/cgit/openstack-infra/zuul/tree/tests/unit/test_scheduler.py?h=feature/zuulv3#n2703 | 20:55 |
| jeblair | adam_g: yeah, that custom function only adds the tags to a build variable. i think instead we should add it to the ansible variables that are available to the job (so, inside launcher/client.py). | 20:59 |
| Shrews | jeblair: good point. i think i'll just combine 429848 into 428428 since i'm going to address your comments with another patch set | 20:59 |
| Shrews | and add INIT there | 20:59 |
| jeblair | Shrews: sounds good. also, INIT is way better than my suggestion. | 20:59 |
| Shrews | lol | 21:00 |
| jeblair | adam_g: i think i wrote it that way originally because we needed it quickly and i didn't want to extend the surface area of build environmental variables. ansible vars makes that far less worrisome. | 21:00 |
| Shrews | jeblair: i'm just glad the bulk of that launch manager craziness was understandable | 21:02 |
| jeblair | Shrews: yeah, that makes a lot of sense. | 21:03 |
| Shrews | just seemed to be the most logical way to treat bulk launches as a single thing | 21:03 |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul feature/zuulv3: Re-enable test_queue_names https://review.openstack.org/429883 | 21:26 |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul feature/zuulv3: Re-enable test_queue_names https://review.openstack.org/429883 | 21:29 |
| rcarrillocruz | clarkb, mordred : https://review.openstack.org/#/c/403732/13 seems to work, i ssh'd to a node running the gate tempst job, .pydistutils gets installed on tempest, root and stack home folders | 21:45 |
| rcarrillocruz | and with that, afk, kid sick | 21:46 |
| clarkb | rcarrillocruz: I applied my +2 | 21:46 |
| rcarrillocruz | saw it thx, jhesketh if you could +A that'd be grand | 21:46 |
| mordred | jhesketh: I +2's but left off my +A in case you wanted to check out the follow up | 21:49 |
| Shrews | grr... anyone have any idea why a nodepool DBTestCase derived test would hang in the teardown portion? | 21:56 |
| jeblair | Shrews: are we still doing any mysql stuff or is that all gone? | 21:56 |
| Shrews | jeblair: should be all gone | 21:57 |
| * SpamapS pours one out for the DB | 21:58 | |
| jeblair | Shrews: hrm... i think i still see it there on my feature/zuulv3 branch... | 21:58 |
| Shrews | jeblair: this is a new test (test_nodelaunchmanager). i changed it to derive from DBTestCase so I could pass the zk object to the launch thread | 21:58 |
| jhesketh | Morning | 21:58 |
| Shrews | jhesketh: by "gone" i meant, not using it | 21:59 |
| Shrews | err, jeblair ^^ | 21:59 |
| Shrews | jeblair: we still need to remove setup stuff | 21:59 |
| jeblair | Shrews: ah... well that fixture has problems and absolutely could cause it to hang during teardown... if we can go ahead and remove it from dbtestcase, that would be swell... | 21:59 |
| mordred | ++ | 22:00 |
| Shrews | jeblair: yep, on the TODO list | 22:00 |
| jeblair | Shrews: however, that should be an intermittent error -- if you're consistently seeing errors, might be something else. | 22:00 |
| Shrews | anyway, meeting | 22:00 |
| jeblair | (tbh, i don't know if the *fixture* has problems or mysql does.... but either way...) | 22:00 |
| jeblair | it's zuul meeting time in #openstack-meeting-alt | 22:00 |
| *** hashar has quit IRC | 22:12 | |
| pabelanger | EOD for today. I'll get started on servers first thing tomorrow for PTG | 23:01 |
| jeblair | pabelanger: thanks! | 23:01 |
| jeblair | i'll paste the ptg prep link here for easy recall: https://etherpad.openstack.org/p/pike-ptg-zuul | 23:02 |
| Shrews | jeblair: maybe a shortened url for the /topic ? | 23:02 |
| Shrews | eh, we have too many links there now | 23:03 |
| ianw | did zuulv3 decide against tagging -rc type releases for any particular reason, or it just wasn't needed? | 23:05 |
| ianw | tagging on the v3 branch i mean | 23:05 |
| ianw | i'm wondering if i tag a rc on the dib v2 branch if it causes more problems than it helps | 23:05 |
| mordred | ianw: it should work - we have support for a pre-release pipeline - but I can't think of any instances of us having ever made pre-release releases of infra related things | 23:06 |
| ianw | heh, it's the should bit that worries me :) | 23:08 |
| mordred | :) | 23:09 |
| mordred | ianw: what outcome are you trying to get to and/or avoid | 23:09 |
| mordred | ? | 23:09 |
| ianw | me tagging a v2 rc release and everything like nodepool, triple-o and other odd users deciding it's the thing they want to use right now | 23:10 |
| ianw | although, most of those should be using dib from upper-constraints, so that lowers risk | 23:10 |
| mordred | well - pip install will also skip the rc release by default | 23:12 |
| mordred | so anyone installing from pip will not get the rc release unless they explicitly request it | 23:12 |
| mordred | _that_ part I'm confident in | 23:13 |
| clarkb | (whcih is why it would be nice for ansible to do that too and then we can avoid weird install dances for security updates) | 23:16 |
| mordred | ++ | 23:18 |
| ianw | ok, i think i'll take my chances on the "should" :) | 23:21 |
| ianw | pabelanger -- would be nice when you're setting up bits to use at the PTG if any dib bits use the v2 branch | 23:21 |
| ianw | i will be sending a detailed announcement about it soon | 23:22 |
| openstackgerrit | Monty Taylor proposed openstack-infra/nodepool master: Consume Task and TaskManager from shade https://review.openstack.org/414759 | 23:26 |
| *** pleia2_ has joined #zuul | 23:29 | |
| *** mmedvede_ has joined #zuul | 23:33 | |
| *** pleia2 has quit IRC | 23:34 | |
| *** mordred has quit IRC | 23:34 | |
| *** mmedvede has quit IRC | 23:34 | |
| *** olaph has quit IRC | 23:34 | |
| *** nibalizer has quit IRC | 23:34 | |
| *** mmedvede_ is now known as mmedvede | 23:35 | |
| *** mordred has joined #zuul | 23:40 | |
| *** olaph has joined #zuul | 23:41 | |
| *** nibalizer has joined #zuul | 23:42 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!