| clarkb | that isn't a regression right? | 00:01 |
|---|---|---|
| clarkb | it has always had that issue I expect | 00:01 |
| tristanC | clarkb: it's not a regression indeed, it just prevent adding the log-inventory role to job that doesn't have a nodeset | 00:02 |
| clarkb | tristanC: but it was already there just in a different role? | 00:03 |
| ianw | clarkb: it is actually "new" behaviour after they re-factored all the inventory scanning logic in 8f97aef1a365cbbbb822d6d09f96af17a076b295 | 00:05 |
| tristanC | clarkb: yes, it's just that we didn't used the validate-host role for job running on the executor, and i just found out the issue when refactoring all base jobs to include the log-inventory role | 00:05 |
| clarkb | ah | 00:05 |
| clarkb | ianw: gerrit doesn't seem to know that sha1 | 00:06 |
| clarkb | ansible maybe? | 00:07 |
| tristanC | it sounds like zuul should set that variable none-the-less, otherwise job have to manually define it using /tmp/{{ zuul.build }}/ansible/inventory.yaml | 00:07 |
| ianw | is it always "/tmp" ? | 00:08 |
| tristanC | ianw: i'm not sure, this just happens to work :-) | 00:09 |
| tristanC | though it's more of a hack until we figure out if/how zuul can set this variable for nodeless jobs | 00:10 |
| clarkb | and I guess in the nodelss job case we want the data that isn't node data as much as the git information and similar | 00:11 |
| ianw | i'm not sure, ansible has made the choice to not populate it in that case | 00:11 |
| tristanC | clarkb: exactly, a dump of zuul dict is rather useful | 00:11 |
| ianw | http://docs.ansible.com/ansible/latest/porting_guide_2.4.html#inventory | 00:11 |
| ianw | so, IMO, it probably makes sense for zuul-jobs level to copy it, but zuul itself to stick closer to what ansible does. but i imagine opinions vary :) | 00:12 |
| *** yolanda_ has quit IRC | 00:16 | |
| openstackgerrit | Merged openstack-infra/zuul master: Change test prints to log.info https://review.openstack.org/554058 | 00:24 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul-jobs master: log-inventory: set inventory_file if needed https://review.openstack.org/578234 | 00:25 |
| tristanC | ianw: 578234 should work as a zuul-jobs level fix | 00:25 |
| ianw | tristanC: yeah, was just digging in to the /tmp issue ... | 00:25 |
| tristanC | zuul uses https://docs.python.org/3/library/tempfile.html#tempfile.gettempdir which should be addressed by 578234 | 00:26 |
| ianw | zuul/executor/client.py execute() i wonder if we should put the actual path in there | 00:26 |
| tristanC | i didn't check for '/tmp' presence since bwrap automatically creates it | 00:27 |
| tristanC | wait no, this can be configured with [executor] job_dir | 00:29 |
| corvus | tristanC: perhaps we should add another variable for zuul.executor.inventory https://zuul-ci.org/docs/zuul/user/jobs.html#id1 | 00:31 |
| tristanC | so either zuul pass the job_dir variable to the job, or either we could re-use {{ zuul.executor.work_root }}/../ansible/ | 00:31 |
| corvus | we've moved the inventory several times to date, so i'd feel better with an explicit variable rather than relying on relative paths | 00:32 |
| tristanC | ok, and then we replace inventory_file with zuul.executor.inventory | 00:32 |
| corvus | ya | 00:33 |
| corvus | tristanC: do you know if anything/anyone is using log-inventory yet? | 00:34 |
| corvus | tristanC: if so, we may need to wait a bit due to this: https://zuul-ci.org/docs/zuul-jobs/deprecation.html#new-zuul-features | 00:35 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul master: executor: add inventory path to the zuul.executor job variable https://review.openstack.org/578235 | 00:36 |
| corvus | but if no one is using it, we can probably skip or abbreviate that (since it's not a syntax change, it'll just be an undefined variable if anyone uses it) | 00:36 |
| clarkb | I don't think anyone is using it yet unless spamaps is | 00:36 |
| tristanC | corvus: i wasn't able to integrate as-is in software-factory base jobs | 00:37 |
| corvus | tristanC: i'm ambivalent about a release note about that, but it does need docs :) | 00:37 |
| corvus | tristanC: also, maybe we should call it 'inventory_file' to match ansible | 00:38 |
| corvus | (and it's probably a more descriptive name anyway) | 00:38 |
| tristanC | well here is how i was able to make it work: https://softwarefactory-project.io/r/#/c/12523/11/ansible/roles/sf-repos/templates/config/playbooks/openshift/pre.yaml.j2 | 00:38 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul master: executor: add inventory_file path to the zuul.executor job variable https://review.openstack.org/578235 | 00:39 |
| openstackgerrit | Ian Wienand proposed openstack-infra/zuul master: Export inventory_file from zuul.executor https://review.openstack.org/578237 | 00:40 |
| *** Shrews has quit IRC | 00:40 | |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul-jobs master: log-inventory: set inventory_file if needed https://review.openstack.org/578234 | 00:40 |
| ianw | jeez, you look away from irc for two minutes :) | 00:41 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul-jobs master: log-inventory: remove inventory_file usage https://review.openstack.org/578238 | 00:41 |
| tristanC | corvus: 578234 should be backward compatible, then we can land 578238 after a grace period | 00:42 |
| corvus | tristanC: yeah, i guess it will work exactly as well as it does today, won't it? :) | 00:42 |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul master: WIP: Support skip_child_jobs via zuul_return https://review.openstack.org/578230 | 00:44 |
| tristanC | yes, if inventory_file is undefined, then you need 578235 | 00:44 |
| corvus | tristanC: i think adding the comment in 578238 like ianw suggested would be a good idea | 00:44 |
| corvus | (the comment was left on 578235) | 00:45 |
| corvus | oh, er, i guess also adding the comment to zuul would be good. that might be a good comment for the docs. | 00:45 |
| *** Shrews has joined #zuul | 00:46 | |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul-jobs master: log-inventory: remove inventory_file usage https://review.openstack.org/578238 | 00:48 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Revert "Hide queue headers for empty queues when filtering" https://review.openstack.org/578241 | 00:52 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Revert "Upgrade from angularjs (v1) to angular (v6)" https://review.openstack.org/578242 | 00:52 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul master: executor: add inventory_file path to the zuul.executor job variable https://review.openstack.org/578235 | 00:52 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: WIP: Don't add implied branch matchers to project-pipeline variants https://review.openstack.org/577881 | 00:54 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Revert "Upgrade from angularjs (v1) to angular (v6)" https://review.openstack.org/578242 | 00:55 |
| tristanC | corvus: maybe the zuul.openstack.org ui could be fixed by changing the rewrite rule to uses '/t/status.html' instead? | 00:55 |
| tristanC | because of https://git.zuul-ci.org/cgit/zuul/tree/web/util.ts#n17 | 00:56 |
| tristanC | or building it using ZUUL_BASE_HREF=https://zuul.openstack.org/ | 00:57 |
| corvus | tristanC: aiui, that method is supposed to handle both cases, so if it's broken, we should fix that? | 00:58 |
| clarkb | ya see my debugging in -infra tldr is we give jquery getJSON() a path for the URL it appends to https and GETs I think | 00:58 |
| clarkb | but we need to give it a url | 00:58 |
| corvus | tristanC: and building with configuration is explicitly something i want to avoid -- we should be able to distribute a pre-built package that people can use unless they have special requirements. | 00:58 |
| clarkb | google indicates window.location.href rather than window.location.pathname is what we need | 00:59 |
| tristanC | corvus: sure, i meant this could be a band-aid until a fix land instead of reverting | 00:59 |
| *** Shrews has quit IRC | 01:00 | |
| corvus | tristanC: ah yeah. though we'll have broken other folks too. | 01:00 |
| corvus | this won't be the first 'revert revert revert' patch in zuul :) | 01:01 |
| clarkb | using firefox debug console on a random page pprint(window.location.pathname) confirms my suspiecions | 01:02 |
| clarkb | window.location.href is what we want | 01:02 |
| *** Shrews has joined #zuul | 01:02 | |
| corvus | clarkb: want to push up a fix? | 01:03 |
| clarkb | I'm working on start, these helper functons touch enough stuff I think it may want .pathname in some cases | 01:03 |
| tristanC | clarkb: indeed, i didn't notice this fixes was using it: https://review.openstack.org/#/c/573494/3/web/zuul/zuul.service.ts | 01:04 |
| tristanC | and PS40 was also using pathname: https://review.openstack.org/#/c/551989/40/web/zuul/zuul.service.ts | 01:05 |
| openstackgerrit | Clark Boylan proposed openstack-infra/zuul master: Use window.location.href for the href not pathname https://review.openstack.org/578243 | 01:06 |
| clarkb | I don't know that this fix is complete | 01:07 |
| clarkb | and the method name is kind of bad since you can't get an href from a path | 01:07 |
| clarkb | given that we'll need to untangle it a revert might be best, incorporate fixes and try again later? | 01:07 |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul master: WIP: Support skip_child_jobs via zuul_return https://review.openstack.org/578230 | 01:08 |
| *** Shrews has quit IRC | 01:08 | |
| tristanC | clarkb: it is confusing indeed, +1 for the revert | 01:08 |
| clarkb | corvus: just the one change for the revert right? | 01:10 |
| clarkb | I have approved it | 01:11 |
| corvus | yeah, i squashed the other dep into it | 01:11 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul master: web: fix status page flickering https://review.openstack.org/578226 | 01:11 |
| clarkb | we have multiple implementations of getBaseHrefFromPath too | 01:12 |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul master: WIP: Support skip_child_jobs via zuul_return https://review.openstack.org/578230 | 01:12 |
| clarkb | I think the clenaup will be to reduce to one implementation. Rename it to something that makes esnse. Possibly add a new function for getting the basepath if we need that anywhere (I think we do) | 01:13 |
| *** Shrews has joined #zuul | 01:14 | |
| pabelanger | corvus: clarkb: mordred: one thing with still doing the skipped result for zuul.child_jobs compare, I think we also need to all skipped jobs not to result in -1 for the build. I'm unsure if we still want to use SKIPPED as the result or maybe some new value (?) to allow +1 | 01:16 |
| pabelanger | https://review.openstack.org/578230 should show an example of the -1 still | 01:16 |
| *** xinliang_ has joined #zuul | 01:28 | |
| openstackgerrit | Merged openstack-infra/zuul master: Revert "Upgrade from angularjs (v1) to angular (v6)" https://review.openstack.org/578242 | 01:38 |
| *** Shrews has quit IRC | 01:42 | |
| *** Shrews has joined #zuul | 01:44 | |
| *** Shrews has quit IRC | 02:34 | |
| *** Shrews has joined #zuul | 02:41 | |
| openstackgerrit | Xinliang Liu proposed openstack-infra/zuul-jobs master: Use debian-security mirror https://review.openstack.org/578259 | 03:09 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul master: executor: add inventory_file path to the zuul.executor job variable https://review.openstack.org/578235 | 03:29 |
| *** nchakrab_ has joined #zuul | 03:43 | |
| *** nchakrab_ has quit IRC | 03:59 | |
| *** bhavik1 has joined #zuul | 04:05 | |
| *** Shrews has quit IRC | 04:13 | |
| *** Shrews has joined #zuul | 04:14 | |
| *** bhavik1 has quit IRC | 04:22 | |
| *** nguyenhai has quit IRC | 04:26 | |
| openstackgerrit | Merged openstack-infra/zuul master: Don't reconfigure tenant on unprotected branch pushes https://review.openstack.org/576538 | 04:26 |
| openstackgerrit | Merged openstack-infra/zuul master: Fix branch parsing in _event_pull https://review.openstack.org/577371 | 04:26 |
| *** nchakrab has joined #zuul | 04:33 | |
| *** nguyenhai has joined #zuul | 04:33 | |
| *** nguyenhai_ has joined #zuul | 04:33 | |
| *** nguyenhai has quit IRC | 04:34 | |
| *** nguyenhai_ has quit IRC | 04:34 | |
| *** nguyenhai has joined #zuul | 04:34 | |
| *** hwoarang has quit IRC | 04:40 | |
| *** Shrews has quit IRC | 04:53 | |
| *** Shrews has joined #zuul | 04:55 | |
| *** hwoarang has joined #zuul | 05:23 | |
| *** Shrews has quit IRC | 05:46 | |
| *** yolanda_ has joined #zuul | 05:47 | |
| *** Shrews has joined #zuul | 05:48 | |
| tristanC | i'm seing a weird behavior since we updated to latest nodepool master, we have a provider with multiple pools and different labels list... | 05:59 |
| tristanC | and node request get assigned to pool without the requested label, and the handler raise a KeyError in the imagesAvailable method() | 06:00 |
| *** hwoarang has quit IRC | 06:06 | |
| *** yolanda__ has joined #zuul | 06:06 | |
| *** hwoarang has joined #zuul | 06:06 | |
| *** Rohaan has joined #zuul | 06:09 | |
| *** yolanda_ has quit IRC | 06:09 | |
| *** bhavik1 has joined #zuul | 06:09 | |
| *** yolanda__ has quit IRC | 06:14 | |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/nodepool master: openstack: decline request for labels un-available in current pool https://review.openstack.org/578288 | 06:19 |
| *** threestrands has quit IRC | 06:19 | |
| tristanC | Shrews: Is is supported to have multiple pools with different labels list? (cc pabelanger ^) | 06:20 |
| tristanC | well that KeyError (silenced in 578288) is also happening in the test_multiple_pools test, and in our case we have 3 pools with 3 different labels list, which make quite a mess when nodepool keeps on declining request assigned to the wrong pool | 06:28 |
| *** hashar has joined #zuul | 06:28 | |
| *** dmellado has joined #zuul | 06:29 | |
| *** yolanda__ has joined #zuul | 06:30 | |
| tristanC | i wonder if it starts hapenning with the new getSupportedLabels logic | 06:34 |
| tristanC | and perhaps this logic needs to be refactor so that we don't assign request to pool when the labels aren't available in the first place | 06:39 |
| *** pcaruana has joined #zuul | 06:44 | |
| *** yolanda__ has quit IRC | 06:50 | |
| *** yolanda__ has joined #zuul | 06:53 | |
| *** yolanda__ is now known as yolanda | 06:53 | |
| *** gtema has joined #zuul | 06:57 | |
| *** pcaruana has quit IRC | 07:02 | |
| *** Shrews has quit IRC | 07:29 | |
| *** Shrews has joined #zuul | 07:31 | |
| *** nguyenhai has quit IRC | 07:43 | |
| *** jpena|off is now known as jpena | 07:46 | |
| *** nguyenhai has joined #zuul | 07:48 | |
| *** pcaruana has joined #zuul | 07:57 | |
| *** bhavik1 has quit IRC | 08:29 | |
| *** electrofelix has joined #zuul | 08:38 | |
| *** ianychoi has quit IRC | 09:14 | |
| *** ianychoi has joined #zuul | 09:15 | |
| *** jimi|ansible has quit IRC | 09:27 | |
| *** threestrands has joined #zuul | 09:45 | |
| *** threestrands has quit IRC | 09:45 | |
| *** threestrands has joined #zuul | 09:45 | |
| *** threestrands has quit IRC | 09:46 | |
| *** threestrands has joined #zuul | 09:47 | |
| *** threestrands has quit IRC | 09:47 | |
| *** threestrands has joined #zuul | 09:47 | |
| *** Rohaan has quit IRC | 10:12 | |
| *** pcaruana has quit IRC | 10:31 | |
| *** Rohaan has joined #zuul | 11:07 | |
| *** jpena is now known as jpena|lunch | 11:15 | |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul master: Add min_avail_hdd governor for zuul-executor https://review.openstack.org/577948 | 11:26 |
| pabelanger | tobiash: ^thanks, updated | 11:27 |
| pabelanger | tristanC: Shrews: I would expect it to be supported I suspect we hit a bug since we only have a single pool in openstack-infra nodepool now | 11:28 |
| *** xinliang_ has quit IRC | 11:38 | |
| openstackgerrit | Merged openstack-infra/zuul-jobs master: Use debian-security mirror https://review.openstack.org/578259 | 11:47 |
| *** Shrews has quit IRC | 11:48 | |
| *** Shrews has joined #zuul | 11:50 | |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul master: Add zuul.child_jobs in ansible inventory file https://review.openstack.org/578181 | 12:04 |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul master: WIP: Support skip_child_jobs via zuul_return https://review.openstack.org/578230 | 12:04 |
| pabelanger | tobiash: ^and reno update | 12:04 |
| *** jpena|lunch is now known as jpena | 12:08 | |
| *** ttx has quit IRC | 12:09 | |
| *** jimi|ansible has joined #zuul | 12:21 | |
| *** jimi|ansible has joined #zuul | 12:21 | |
| *** ttx has joined #zuul | 12:22 | |
| *** rlandy has joined #zuul | 12:29 | |
| *** myoung|off is now known as myoung | 12:47 | |
| *** nchakrab_ has joined #zuul | 12:58 | |
| *** hashar is now known as idoine | 12:59 | |
| *** Shrews has quit IRC | 13:00 | |
| *** nchakrab has quit IRC | 13:00 | |
| electrofelix | does zuulv3 gate different branches of a project separately automatically? or does this always require separate pipelines to be explicitly defined. | 13:04 |
| *** Shrews has joined #zuul | 13:07 | |
| *** Rohaan has quit IRC | 13:31 | |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul master: Use window.location.href for the href not pathname https://review.openstack.org/578243 | 13:37 |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul master: web: fix status page flickering https://review.openstack.org/578226 | 13:37 |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul master: Revert "Revert "Upgrade from angularjs (v1) to angular (v6)"" https://review.openstack.org/578384 | 13:37 |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul master: Consolidate navigation functions https://review.openstack.org/578385 | 13:37 |
| mordred | electrofelix: automatically, just like zuul v2 | 13:38 |
| electrofelix | mordred: so all branches gated together by default | 13:38 |
| mordred | electrofelix: yeah. I mean - patches submitted trigger jobs for all branches by default. jobs can choose to limit which branches they run on (matchers) also, in-repo .zuul.yaml files are read from all branches and content found in a branch has an implied branch-matcher added to it | 13:40 |
| mordred | and then if you have a multi-repo job, zuul will, by default, get matching branches if possible - so a patch to stable/newton of nova that triggers a job that needs nova and neutron would also work with stable/newton of neutron if it exists | 13:41 |
| mordred | electrofelix: which I *think* is alongwinded way of saying "yes" | 13:42 |
| electrofelix | I presume this also means there is no easy way to have zuul automatically gate branches stable/<x> separately to master where you might not know stable/<x> beforehand? | 13:48 |
| mordred | electrofelix: I don't know what you mean by that? | 13:52 |
| mordred | electrofelix: you don't have to do anything specific to gate branches stable/<x> ... if you only have one definition of a job and it's in master it will apply to all branches (unless you add a specific branch matchers) - however, if you put a .zuul.yaml into stable/<x> branch, then that file will automatically DTRT and be a description of what jobs you want applied to that branch | 13:53 |
| mordred | electrofelix: (It's possible i'm saying things poorly here) | 13:53 |
| electrofelix | just meant that 2 changes to the same project for separate branches get gated together unless you explicitly create a separate pipeline for each branch | 13:55 |
| mordred | electrofelix: well, they'll both go into the gate pipeline- but I'm not sure what 'gated together' means in this context? we definitely do not have branch-specific-pipelines in openstack | 13:57 |
| clarkb | I think the ask is for them to be in different queues in the gate so stable failing doesnt affect master and vice versa | 13:58 |
| electrofelix | mordred: maybe we've just configured something wrong, they aren't quite gated together but it appears that a change to stable/x will end up waiting for a change to master for the same project to be completed even though it's independent because it doesn't affect the master branch | 13:58 |
| clarkb | which prevents you from doing upgrade testing which is why that isnt the behavior you get out of the box | 13:58 |
| clarkb | if you set a different name on each branch explicitly I think that should produce the desired behavior | 14:00 |
| clarkb | but am not aware of any other option for that | 14:00 |
| electrofelix | clarkb: thanks, we're definitely dealing with a bit of an edge case here | 14:02 |
| corvus | the dependent pipeline manager groups queues by project; i don't believe there's a way to have it do so by project-branch | 14:04 |
| mordred | ah. I grok what we're talking about now - sorry | 14:05 |
| corvus | https://zuul-ci.org/docs/zuul/user/config.html#attr-project.%3Cpipeline%3E.queue | 14:05 |
| corvus | that says that the first queue that appears for a project-pipeline wins. so if the value is different on different branches, zuul will use the first one (usually master) | 14:06 |
| *** Shrews has quit IRC | 14:06 | |
| clarkb | ah | 14:06 |
| *** Shrews has joined #zuul | 14:08 | |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul master: Consolidate navigation functions https://review.openstack.org/578385 | 14:10 |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul master: web: fix status page flickering https://review.openstack.org/578226 | 14:10 |
| mordred | clarkb: ^^ 578385 takes care of the thing you were mentioning about there being multiple copies of those functions | 14:11 |
| clarkb | mordred: is getBaseHrefFromPath() an accurate name now that we use window.location.href instead of window.location.pathname? | 14:12 |
| mordred | clarkb: it's not - also - I don't think that change works yet | 14:13 |
| clarkb | mordred: the slice at the end of https://review.openstack.org/#/c/578385/2/web/util.ts uses the wrong end index. I think you want -2 ? | 14:14 |
| clarkb | -1 sorry | 14:15 |
| * clarkb leaves a comment | 14:15 | |
| mordred | yes | 14:15 |
| *** Shrews has quit IRC | 14:17 | |
| *** nchakrab_ has quit IRC | 14:18 | |
| clarkb | mordred: my bigget concern with just trying to jam my fix in was I think we may have used the path munging in a place or two | 14:23 |
| clarkb | so getHreffromPath was really getPathPrefix, but I wasn't super sure on that | 14:24 |
| mordred | yah - I think the revert was the right choice so we can go back through and make sure it's right in all the places | 14:29 |
| *** threestrands has quit IRC | 14:30 | |
| *** Shrews has joined #zuul | 14:32 | |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul master: Consolidate navigation functions https://review.openstack.org/578385 | 14:32 |
| mordred | clarkb: ok - that works for openstack and software-factory in local testing | 14:32 |
| mordred | clarkb: but I need to do a test without ZUUL_API_URL set still | 14:33 |
| clarkb | mordred: looks like latest patchset will result in the href always ending in / | 14:35 |
| clarkb | but previous attempted to remove a / if it was there | 14:35 |
| clarkb | / ~= / so should be fine, but that may need cleaning up in some places if it expected no / | 14:35 |
| mordred | yah - I put that in just in this patch to make it more similar to what getBaseHrefFromPath - but it turns out it breaks things | 14:36 |
| clarkb | heh irc // esaped to one / | 14:36 |
| clarkb | I meant // ~= / | 14:36 |
| *** acozine1 has joined #zuul | 14:53 | |
| *** acozine1 has quit IRC | 14:54 | |
| *** acozine1 has joined #zuul | 15:03 | |
| *** acozine1 has quit IRC | 15:17 | |
| pabelanger | Did i see a patch to nodepool recently to allow nodepool to create tenant networks for nodesets? | 15:56 |
| pabelanger | maybe it was some discussion | 15:56 |
| *** myoung is now known as myoung|biaf | 15:59 | |
| openstackgerrit | Artem Goncharov proposed openstack-infra/nodepool master: retire shade in favor of openstacksdk https://review.openstack.org/572829 | 16:00 |
| gtema | mordred: I have turned nodepool caching back, but left commented code, how it should be | 16:01 |
| mordred | gtema: awesome - thanks! | 16:05 |
| *** myoung|biaf is now known as myoung | 16:06 | |
| *** idoine is now known as hashar | 16:07 | |
| *** tobasco is now known as tobasco_afk | 16:12 | |
| Shrews | pabelanger: i think there was talk, but no actual code | 16:18 |
| pabelanger | Shrews: okay thanks, I'll look back at eavesdrop.o.o in a bit | 16:20 |
| pabelanger | I know tripleo is using a lot of provider networks for their te-broker stuff, so would be interesting to discuss a way in the near future to delete some of that stuff. | 16:21 |
| clarkb | SpamapS: (I think it was SpamapS) was tlaking about nodepool as becoming "thingpool" | 16:23 |
| corvus | i think we might want to work on the container stuff first. the spec has us adding a non-node resource. i think we can see how that goes before we add more thingpool (not to mention, we have a bunch of drivers waiting in the wings which we should attend to after that) | 16:23 |
| clarkb | pabelanger: you could have them use overlays like we do for multinode testing | 16:23 |
| clarkb | (as a today option) | 16:24 |
| clarkb | corvus: ++ | 16:24 |
| pabelanger | clarkb: yah, that is what I am hope to change them to first | 16:25 |
| corvus | good. i hope we get there eventually (i think thingpool sounds like a good idea), but i'm glad there are other options for now, cause i don't think we'll be ready very soon. | 16:27 |
| *** hashar is now known as hasharAway | 16:28 | |
| pabelanger | corvus: clarkb: tobiash: could I ask you to add https://review.openstack.org/577948/ so your review pipeline, we've added another zuul-executor to rdo yesterday, but still think it would be helpful. | 16:31 |
| clarkb | pabelanger: left some comments, -1 due to what I am pretty sure is bug in stats reporting | 16:41 |
| *** rlandy is now known as rlandy|brb | 16:45 | |
| *** ssbarnea has left #zuul | 16:47 | |
| corvus | pabelanger, clarkb: statsd/graphite deal with integers, so the actual statsd value recorded is "hundredths of a percent". that way we can report the percentage to 2 decimal places | 16:48 |
| corvus | probably the bug there is the lack of comment | 16:48 |
| clarkb | ah | 16:49 |
| corvus | the documentation *does* mention that, however. but pabelanger's patch doesn't, so that at least needs to be updated. | 16:51 |
| mordred | clarkb, corvus: fwiw, I'm off in the corner bludgeoning myself with javascript - so if I don't respond quickly, well, just wish me luck :) | 16:52 |
| corvus | mordred: i figured as much and am sending you thoughts of fluffy javascript bunnies | 16:52 |
| *** myoung is now known as myoung|lunch | 17:02 | |
| *** rlandy|brb is now known as rlandy | 17:07 | |
| *** jpena is now known as jpena|off | 17:12 | |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Don't add implied branch matchers to project-pipeline variants https://review.openstack.org/577881 | 17:44 |
| *** electrofelix has quit IRC | 17:46 | |
| *** gtema has quit IRC | 18:04 | |
| *** yolanda_ has joined #zuul | 18:11 | |
| *** Shrews has quit IRC | 18:14 | |
| *** yolanda has quit IRC | 18:14 | |
| *** Shrews has joined #zuul | 18:15 | |
| *** myoung|lunch is now known as myoung | 18:21 | |
| *** Shrews has quit IRC | 18:28 | |
| *** Shrews has joined #zuul | 18:29 | |
| corvus | fungi made a suggestion in #openstack-infra which we may want to explore at some point: have zuul perform a kind of dynamic configuration on items without branches to figure out their "branced" config. in other words, a way to have branch-specific jobs for tag/release based pipelines. | 18:31 |
| corvus | i think it would be possible, if a bit resource intensive (since i think you would have to perform that configuration on every item in those pipelines) | 18:32 |
| *** Shrews has quit IRC | 18:33 | |
| *** Shrews has joined #zuul | 18:33 | |
| Shrews | freenode hates me | 18:34 |
| fungi | yeah, i kinda expected it wouldn't be a slam dunk feature | 18:35 |
| fungi | was more brainstorming, as the surrounding discussion was about branched configuration not percolating into tags for commits on their respective branches | 18:36 |
| fungi | and acknowledging that tags don't necessarily map to one (and only one) branch, but that we do potentially have configuration inside the repository state that tag represents | 18:37 |
| fungi | similar to speculative configuration testing on proposed changes/pull requests | 18:38 |
| corvus | yeah, trying to identify the branch a tag is on is something we could have done in v2 but did not because the answer could be ambiguous (a tagged commit could be on zero, one, or more branches) | 18:39 |
| corvus | the v3 suggestion of dynamic config is unambiguous at least as far as in-repo config goes. it wouldn't let you, say, centrally define a project-pipeline which says "run foo on releases from stable and bar on releases from master". | 18:40 |
| corvus | so even that isn't a 100% solution to the "problem" | 18:40 |
| *** rlandy is now known as rlandy|afk | 18:41 | |
| *** tobasco_afk is now known as tobasco | 18:44 | |
| fungi | right, explicit branch matchers wouldn't work, but implicit could still | 18:46 |
| fungi | or rather, explicit branch matchers work today you just (as you pointed out) get a pattern-match on the tag name itself | 18:46 |
| fungi | which in most cases isn't much help | 18:47 |
| fungi | and i guess the implicit branch matcher in this case wouldn't be the normal one, but would be more akin to dynamic configuration in proposed changes | 18:49 |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul master: Add min_avail_hdd governor for zuul-executor https://review.openstack.org/577948 | 18:50 |
| pabelanger | corvus: clarkb: okay, I've updated docs based on comments^ | 18:51 |
| pabelanger | I'll look into a constant for /var/lib/zuul now | 18:51 |
| corvus | fungi: right | 18:52 |
| *** hasharAway is now known as hashar | 19:09 | |
| dmsimard | I'd really like to add a role to the role path of the zuul executor and that role has nothing to do with zuul, is it not possible to have it available without jumping through the hoops of adding the role to the zuul config and adding it to the roles stanza ? The executor prohibits running local code so I can't clone or otherwise retrieve it "manually" | 19:16 |
| pabelanger | dmsimard: I believe that is the idea with https://zuul-ci.org/docs/zuul/user/config.html#attr-job.roles.galaxy but don't think it is implemented yet. Give that galaxy is just github.com, I wonder if we should revisit that in the future | 19:19 |
| pabelanger | I've always wanted to see if we could have zuul.jobs point to an ansible-galaxy requirements file too, so avoid duplicating that info in zuul.yaml | 19:20 |
| dmsimard | pabelanger: galaxy is a bit more than a wrapper over git but yeah, maybe a generic git driver | 19:21 |
| *** Shrews has quit IRC | 19:21 | |
| pabelanger | dmsimard: right, IIRC I don't think galaxy has anything that hosts tarballs. It just depends upon github infrastructure for that | 19:22 |
| dmsimard | oh yeah galaxy uses tarballs, it doesn't outright clone | 19:23 |
| dmsimard | at least it was that way when I last looked at the code | 19:23 |
| dmsimard | which is not inherently bad, I guess | 19:23 |
| *** Shrews has joined #zuul | 19:26 | |
| pabelanger | dmsimard: which laptop did you get? | 19:27 |
| pabelanger | dmsimard: wrong windo | 19:28 |
| *** Shrews has quit IRC | 19:44 | |
| *** Shrews has joined #zuul | 19:53 | |
| corvus | clarkb, fungi, SpamapS, tobiash, mordred, tristanC, andreaf, EmilienM: https://review.openstack.org/577881 is a subtle but important change to branch matchers and project-templates. if you have a minute to think about that and/or suggest any use cases to compare against to make sure that it doesn't break anything unexpected, i'd appreciate it. | 20:11 |
| * SpamapS will look at it later today | 20:15 | |
| corvus | oh and i should have said it's not urgent :) | 20:16 |
| tobiash | corvus: will look at it tomorrow when my brain is fresh | 20:18 |
| tobiash | Speaking about reviews, it would be great to get a non-corvus vote on 575014 | 20:22 |
| tobiash | That blocks the github multi enqueue fix currently | 20:22 |
| tobiash | It's a small logging enhancement | 20:23 |
| tobiash | corvus: do you want to have another look onto the hdd governor or shall I +3 it? | 20:26 |
| corvus | tobiash: i'd like another look, i only did a quick pass earlier to clear up questions | 20:26 |
| tobiash | ok | 20:26 |
| *** yolanda_ has quit IRC | 20:43 | |
| *** yolanda_ has joined #zuul | 20:44 | |
| SpamapS | hdd governor? neat | 20:49 |
| *** yolanda has joined #zuul | 20:53 | |
| *** yolanda_ has quit IRC | 20:55 | |
| *** yolanda_ has joined #zuul | 20:59 | |
| *** yolanda has quit IRC | 21:02 | |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: WIP: Match tag items against containing branches https://review.openstack.org/578557 | 21:06 |
| corvus | fungi, clarkb: ^ because i have no self control and enjoy programming, i sketched out a "git branch --contains" approach might look like. | 21:07 |
| fungi | wow | 21:08 |
| fungi | i'll take a look after okonomiyaki are done | 21:08 |
| fungi | already starred your implied matchers change to look at this evening too ;) | 21:08 |
| corvus | the number of lines added is about the same number of lines in the commit message | 21:09 |
| fungi | my review comments will hopefully not be influenced by the cheap sake accompanying the main course | 21:10 |
| corvus | fungi: "this change is full bodied with notes of jasmine"? | 21:11 |
| fungi | more like "fruity mold flavors" | 21:11 |
| corvus | yes that does sound like my code | 21:11 |
| fungi | christine asked if you had a direct between beijing and sfo, and what the flight time is on that | 21:13 |
| fungi | though i assume it differs depending on which direction you're headed | 21:13 |
| corvus | yes, and oddly enough, there was like no wind this time so it was, erm, 11:30? both ways. | 21:14 |
| fungi | oh, neat | 21:14 |
| fungi | not as bad as i imagined | 21:15 |
| fungi | though she thought it would be shorter | 21:15 |
| fungi | hk/newark was a solid 18 hours for me | 21:16 |
| corvus | apparently it's 12:30 scheduled, but 11:30-12:00 are pretty typical times | 21:16 |
| corvus | and 12:00 scheduled back | 21:16 |
| corvus | ua 888 / 889 | 21:17 |
| fungi | i need to get my entry visa process started | 21:20 |
| clarkb | fungi: is this for the TC thing? | 21:21 |
| clarkb | or are you going for fun? | 21:21 |
| fungi | just in general for future events | 21:21 |
| fungi | i didn't go this time mainly because i was concerned about turn around time on the visa as well as just not wanting to travel so much | 21:22 |
| corvus | sfo consulate turnaround is 4 days | 21:22 |
| fungi | having a visa lined up without any future travel plans makes it a little easier for me to work with shorter timelines | 21:23 |
| clarkb | ya I decdied against it and am honeslty glad about it. I've managed to get some stuff down around the house and go boating with my brother and I just wouldn't have time if doing another international trip | 21:23 |
| fungi | but wow, 4 days is lightning | 21:23 |
| corvus | to be fair, i did literally spend 1.5 of those days in the waiting room. it did not feel like lightning at the time. :) | 21:23 |
| clarkb | also lots of grilling | 21:24 |
| fungi | hah | 21:24 |
| openstackgerrit | Merged openstack-infra/zuul master: executor: add inventory_file path to the zuul.executor job variable https://review.openstack.org/578235 | 21:37 |
| openstackgerrit | Merged openstack-infra/zuul master: Add pause function to executor https://review.openstack.org/577827 | 21:40 |
| *** myoung is now known as myoung|pto | 21:41 | |
| corvus | clarkb: trivial +3 https://review.openstack.org/575014 | 21:42 |
| *** hashar has quit IRC | 21:42 | |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: WIP: Match tag items against containing branches https://review.openstack.org/578557 | 21:45 |
| clarkb | corvus: done | 21:47 |
| openstackgerrit | Merged openstack-infra/zuul master: Add min_avail_hdd governor for zuul-executor https://review.openstack.org/577948 | 22:09 |
| *** Shrews has quit IRC | 22:12 | |
| *** rlandy|afk is now known as rlandy | 22:15 | |
| openstackgerrit | Merged openstack-infra/zuul master: Improve logging of GithubTriggerEvents https://review.openstack.org/575014 | 22:16 |
| *** Shrews has joined #zuul | 22:22 | |
| openstackgerrit | Merged openstack-infra/zuul master: Fix missing dequeue on synchronized pull request https://review.openstack.org/577355 | 22:29 |
| tristanC | corvus: could we please have a nodepool 3.0.2 tag? And maybe a zuul 3.1.1 too? | 22:47 |
| clarkb | tristanC: we are doing a nodepool burn in on openstack's eployment I expect a new release will happen soon as I am not aware of any major issues | 22:48 |
| tristanC | out of curiosity, what is the release process, can any core git push --tags? | 22:50 |
| tristanC | clarkb: thanks, that's appreciated :-) | 22:51 |
| clarkb | tristanC: any member of this group can push a tag https://review.openstack.org/#/admin/groups/745,members. You should never push --tags and instead push the specific tag you intend on pushing (othrewise you can push all kinds of refs into gerrit that you never inteded to push) also the tags must be signed | 22:52 |
| clarkb | also double check that the sha1 you tagged is in gerrit or you can push an entire set of commits to gerrit you never intended on pushing | 22:52 |
| tristanC | oh, that's a good advice, then i've been lucky so far | 22:54 |
| fungi | note that whether someone can push a tag is not the same question as asking about the release process | 22:54 |
| fungi | the release process, whether codified or not, is certainly far more than pushing a tag | 22:54 |
| *** Shrews has quit IRC | 22:55 | |
| clarkb | the process zuul and nodepool appear to be adopting is to run the code in openstack (so ensure services are restarted) for a ~day and make sure it works then if so tag it using semver based version. Pushing this tag to gerrit then triggers artifact builds and publishing | 22:56 |
| tristanC | fungi: good point, i meant to ask if there was a repo similar to openstack/release where anyone could propose a new release | 22:56 |
| corvus | tristanC: i thought you reported a problem with nodepool master? | 22:56 |
| tristanC | corvus: i can't tell if it's related to the new getSupportedLabels logic or if it was already affecting 3.0.1 | 22:58 |
| *** Shrews has joined #zuul | 22:58 | |
| corvus | tristanC: right that was the thing i was thinking of | 22:58 |
| corvus | tristanC: can you work with Shrews to figure that out? i'd like to wait until that's answered one way or the other, then i think we can release. i haven't heard about any problems in openstack's deployment, so i think that's clear. | 23:00 |
| fungi | de facto release process also seems to entail posting a release notice to the zuul-announce ml (pointing out any significant changes present in the release, hopefully represented by release notes) | 23:00 |
| tristanC | corvus: did the openstack-infra^Wwinterscale launchers got restarted with the current master? | 23:00 |
| corvus | tristanC: i believe Shrews did that, yes (not winterscale yet) | 23:01 |
| corvus | and i'm happy to continue actually making the releases (since that mostly entails checking with people that things are ready to go, which is something i do anyway) | 23:03 |
| tristanC | Shrews: the KeyError silenced in https://review.openstack.org/#/c/578288/ does not happen in nodepool-3.0.1 | 23:07 |
| tristanC | corvus: oh well, then the nodepool tag will have to wait, I don't think the exception is really an issue, but it makes launcher.log pretty messy... | 23:09 |
| tristanC | ftr, here is our nodepool.yaml: https://softwarefactory-project.io/cgit/config/tree/nodepool/rdo-cloud.yaml | 23:09 |
| *** yolanda__ has joined #zuul | 23:11 | |
| *** Shrews has quit IRC | 23:13 | |
| clarkb | tristanC: are those pools separated so that they can have different max server values? | 23:14 |
| *** Shrews has joined #zuul | 23:15 | |
| *** yolanda_ has quit IRC | 23:15 | |
| tristanC | clarkb: it seems like this is what pabelanger wants to do yes | 23:15 |
| *** ianychoi has quit IRC | 23:16 | |
| tristanC | we used to have different tenant, but it seems better to have a single provider per openstack endpoint to reduce the number of api calls hitting the same services | 23:16 |
| clarkb | tristanC: were the number of api calls a problem? (nodepool should cache fiarly aggressively) | 23:17 |
| *** yolanda__ has quit IRC | 23:18 | |
| *** ianychoi has joined #zuul | 23:19 | |
| tristanC | not from our side, though i assumed this would be appreciated by the rdo-cloud ops to keep api load low | 23:19 |
| *** yolanda has joined #zuul | 23:20 | |
| tristanC | this also makes tenant management easier | 23:20 |
| *** yolanda_ has joined #zuul | 23:21 | |
| corvus | tristanC: yeah, before release i think we need to understand the issue more (why is it only happening now?), and, assuming it's okay, merge your patch to silence the traceback | 23:23 |
| *** yolanda has quit IRC | 23:24 | |
| *** yolanda__ has joined #zuul | 23:25 | |
| tristanC | corvus: the proper fix may be to skip pool assignment early if it doesn't have the needed label | 23:25 |
| tristanC | for example, when you have a static and openstack provider, then each request for a static label may get "declined_by" any openstack pool | 23:26 |
| tristanC | shouldn't the "declined_by" list only describe pool worker that could have actually process the request? | 23:27 |
| tristanC | this get worst when you have many pools with unique labels list | 23:28 |
| *** yolanda_ has quit IRC | 23:29 | |
| *** yolanda has joined #zuul | 23:29 | |
| tristanC | though i'm not sure how much work this refactor would take, and perhaps the proposed KeyError fix is enough for nodepool-3.0.2 | 23:31 |
| *** yolanda__ has quit IRC | 23:31 | |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul master: zk: retry initial zookeeper connection attempts https://review.openstack.org/576048 | 23:45 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul master: zk: use kazoo retry facilities https://review.openstack.org/536209 | 23:45 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/nodepool master: zk: retry initial zookeeper connection attempts https://review.openstack.org/576047 | 23:54 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/nodepool master: zk: use kazoo retry facilities https://review.openstack.org/535537 | 23:54 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!