| *** mattw4 has quit IRC | 00:18 | |
| *** michael-beaver has quit IRC | 00:19 | |
| *** jamesmcarthur has quit IRC | 00:23 | |
| *** sgw has quit IRC | 00:37 | |
| *** noorul has joined #zuul | 00:40 | |
| noorul | ofosos: Looks like canMerge interface got changed in master. It takes extra optional event parameter. | 00:41 |
|---|---|---|
| *** altlogbot_1 has quit IRC | 00:46 | |
| *** altlogbot_1 has joined #zuul | 00:48 | |
| noorul | I am wondering how tests are not finding this out in https://review.opendev.org/#/c/677257/ | 00:49 |
| *** igordc has quit IRC | 01:11 | |
| noorul | corvus: hi | 01:51 |
| noorul | corvus: I am not able to get new log view page | 01:51 |
| noorul | corvus: I added <IfModule mod_headers.c> | 01:51 |
| noorul | Header set Access-Control-Allow-Origin "*" | 01:51 |
| noorul | </IfModule> | 01:51 |
| noorul | to httpd.conf in examples folder | 01:52 |
| noorul | ofosos: It looks like still the merge is failing | 02:13 |
| noorul | ofosos: http://paste.openstack.org/show/760142/ | 02:13 |
| *** noorul has quit IRC | 02:31 | |
| *** noorul has joined #zuul | 03:00 | |
| *** spsurya has joined #zuul | 03:03 | |
| *** bhavikdbavishi has joined #zuul | 03:08 | |
| *** bhavikdbavishi1 has joined #zuul | 03:11 | |
| *** bhavikdbavishi has quit IRC | 03:13 | |
| *** bhavikdbavishi1 is now known as bhavikdbavishi | 03:13 | |
| *** noorul has quit IRC | 03:47 | |
| *** rlandy|ruck|bbl is now known as rlandy|ruck | 03:56 | |
| *** rlandy|ruck has quit IRC | 04:02 | |
| *** raukadah is now known as chkumar|rover | 05:15 | |
| *** sanjayu_ has joined #zuul | 05:32 | |
| *** persia has quit IRC | 05:49 | |
| *** sgw has joined #zuul | 05:52 | |
| openstackgerrit | Merged zuul/zuul master: web: link the buildset page from the build https://review.opendev.org/675493 | 05:57 |
| *** persia has joined #zuul | 06:00 | |
| *** persia has quit IRC | 06:03 | |
| *** persia has joined #zuul | 06:05 | |
| *** persia has quit IRC | 06:05 | |
| *** persia has joined #zuul | 06:05 | |
| *** themroc has joined #zuul | 06:40 | |
| *** jpena|off is now known as jpena | 07:41 | |
| *** ianychoi has quit IRC | 07:46 | |
| *** bjackman has joined #zuul | 07:47 | |
| *** ianychoi has joined #zuul | 07:47 | |
| ofosos | noorul: this can happen, it is usually some misconfiguration | 07:59 |
| ofosos | noorul: this can also happen, when the pr is changed, between the build and the merge. This is what should happen. | 08:01 |
| *** hashar has joined #zuul | 08:08 | |
| *** jangutter has joined #zuul | 08:16 | |
| *** panda has quit IRC | 08:18 | |
| *** panda has joined #zuul | 08:20 | |
| ofosos | noorul: merging works on our installation. Not sure why it doesn't work for you. | 08:27 |
| ofosos | The usual reason for this to happen is a sneak push, that happens between gate and merge. | 08:28 |
| *** jamesmcarthur has joined #zuul | 08:28 | |
| *** yolanda has quit IRC | 08:30 | |
| ofosos | noorul: Also pr_version=0 looks really weird, that should never happen if you have an approve workflow | 08:30 |
| ofosos | Did one of the merges succeed? | 08:32 |
| *** sgw has quit IRC | 08:36 | |
| *** yolanda has joined #zuul | 08:43 | |
| *** jamesmcarthur has quit IRC | 08:57 | |
| *** jamesmcarthur has joined #zuul | 08:59 | |
| *** jamesmcarthur has quit IRC | 09:08 | |
| *** jamesmcarthur has joined #zuul | 09:09 | |
| *** bhavikdbavishi has quit IRC | 09:27 | |
| *** jamesmcarthur has quit IRC | 09:41 | |
| *** sshnaidm|afk is now known as sshnaidm | 09:49 | |
| openstackgerrit | Sorin Sbarnea proposed zuul/zuul-jobs master: DNM: experiment ansible-version: 2.8 https://review.opendev.org/677440 | 10:11 |
| *** jamesmcarthur has joined #zuul | 10:16 | |
| *** jamesmcarthur has quit IRC | 10:23 | |
| *** aluria has quit IRC | 10:42 | |
| *** jamesmcarthur has joined #zuul | 10:55 | |
| *** jamesmcarthur has quit IRC | 11:00 | |
| *** bjackman has quit IRC | 11:22 | |
| *** jamesmcarthur has joined #zuul | 11:36 | |
| *** jamesmcarthur has quit IRC | 11:40 | |
| *** jamesmcarthur has joined #zuul | 11:43 | |
| *** bhavikdbavishi has joined #zuul | 11:51 | |
| *** bhavikdbavishi1 has joined #zuul | 11:54 | |
| *** bhavikdbavishi has quit IRC | 11:55 | |
| *** bhavikdbavishi1 is now known as bhavikdbavishi | 11:55 | |
| *** rlandy has joined #zuul | 12:02 | |
| *** rlandy is now known as rlandy|ruck | 12:03 | |
| *** hashar has quit IRC | 12:16 | |
| *** hashar has joined #zuul | 12:19 | |
| *** jamesmcarthur has quit IRC | 12:24 | |
| *** jamesmcarthur has joined #zuul | 12:25 | |
| *** jamesmcarthur has quit IRC | 12:29 | |
| *** jamesmcarthur has joined #zuul | 12:52 | |
| openstackgerrit | Merged zuul/zuul master: Include ref info in smtp reporter subjects https://review.opendev.org/677285 | 13:08 |
| openstackgerrit | Tristan Cacqueray proposed zuul/zuul master: WIP: builds: set missing start time for merger failure https://review.opendev.org/677466 | 13:09 |
| *** hogepodge has joined #zuul | 13:10 | |
| *** bhavikdbavishi has quit IRC | 13:24 | |
| openstackgerrit | James E. Blair proposed zuul/zuul master: manager: specify report failure in logs https://review.opendev.org/671760 | 13:45 |
| *** michael-beaver has joined #zuul | 14:20 | |
| *** jeliu_ has joined #zuul | 14:27 | |
| openstackgerrit | Merged zuul/zuul master: Switch to review.opendev.org for README.rst https://review.opendev.org/676529 | 14:38 |
| openstackgerrit | Merged zuul/zuul master: Use a requests session to simplify auth'd calls https://review.opendev.org/670511 | 14:44 |
| *** themroc has quit IRC | 14:45 | |
| ofosos | How is the Openshift provider supposed to sync with the executor? | 14:58 |
| *** openstackstatus has quit IRC | 14:58 | |
| *** openstack has joined #zuul | 15:03 | |
| *** ChanServ sets mode: +o openstack | 15:03 | |
| *** mattw4 has joined #zuul | 15:07 | |
| *** chkumar|rover is now known as raukadah | 15:08 | |
| *** electrofelix has joined #zuul | 15:11 | |
| *** noorul has joined #zuul | 15:15 | |
| noorul | ofosos: hi | 15:16 |
| *** bhavikdbavishi has joined #zuul | 15:16 | |
| *** bhavikdbavishi1 has joined #zuul | 15:19 | |
| *** bhavikdbavishi has quit IRC | 15:20 | |
| *** bhavikdbavishi1 is now known as bhavikdbavishi | 15:20 | |
| openstackgerrit | David Shrewsbury proposed zuul/zuul master: Store autohold requests in zookeeper https://review.opendev.org/661114 | 15:20 |
| *** sgw has joined #zuul | 15:25 | |
| *** noorul has quit IRC | 15:34 | |
| *** noorul has joined #zuul | 15:36 | |
| *** dkehn has joined #zuul | 15:36 | |
| ofosos | noorul: did one of the merges succeed? | 15:42 |
| openstackgerrit | Tristan Cacqueray proposed zuul/zuul master: docs: add test jobs howto https://review.opendev.org/676424 | 15:43 |
| openstackgerrit | Tristan Cacqueray proposed zuul/zuul master: docs: add test jobs howto https://review.opendev.org/676424 | 15:46 |
| noorul | I have two PRs both changing the same file. If we merge them one by | 15:48 |
| noorul | one then there will be a conflict. But when I approve these PRs in | 15:48 |
| noorul | parallel, the gate pipeline tests them successfully. I was expecting | 15:48 |
| noorul | it to fail because there is a merge conflict and the second one in | 15:48 |
| noorul | the queue should run with first patch merged together. | 15:48 |
| noorul | This is using bitbucket driver. I am not sure whether a driver has anything to do with this logic | 15:48 |
| corvus | noorul: if git can resolve the conflicts, that might be okay | 15:49 |
| noorul | ofosos: The dependent pipeline is not reporting merge conflict. Looks like it is not merging at all | 15:49 |
| noorul | two PRs changing line 1 of a file | 15:49 |
| corvus | yeah that shouldn't work | 15:50 |
| openstackgerrit | Tristan Cacqueray proposed zuul/zuul-jobs master: DNM: test match-on-config-updates for third-party-CI https://review.opendev.org/677499 | 15:50 |
| noorul | corvus: That is working with bitbucket driver | 15:50 |
| corvus | noorul: it certainly sounds like it could be a bug and should be investigated | 15:51 |
| noorul | corvus: on the driver side? | 15:51 |
| noorul | corvus: where should I look? | 15:52 |
| corvus | noorul: if i were debugging this, i would start by making a test case using zuul's unit testing system. that will let you reproduce the problem, iteratively debug it, verify that it's resolved, and prevent future regressions | 15:54 |
| corvus | it takes a long time to run all the tests, i wouldn't recommend that, but it's pretty easy to run a single test | 15:54 |
| noorul | corvus: hmm | 15:55 |
| corvus | noorul: tests.unit.test_scheduler.TestScheduler.test_project_merge_conflict is an example of a unit test that covers that case. you can adapt it to use the bitbucket driver | 15:55 |
| noorul | corvus: What is the base OS for development ? | 16:01 |
| *** jamesmcarthur has quit IRC | 16:03 | |
| clarkb | noorul: I use opensuse, others use ubuntu and debian and fedora | 16:04 |
| clarkb | The test jobs for zuul itself mostly run on ubuntu and centos iirc | 16:04 |
| noorul | I see python3.5 in tox.ini | 16:04 |
| *** jamesmcarthur has joined #zuul | 16:05 | |
| noorul | Latest Ubuntu has 3.6 I think | 16:05 |
| clarkb | ya it runs against python36 and 37 too (I have 37 locally) | 16:06 |
| noorul | clarkb: oh okay | 16:06 |
| clarkb | we could update the envlist to be more specific about that then tell tox not finding an interpreter version is ok | 16:07 |
| openstackgerrit | Tristan Cacqueray proposed zuul/nodepool master: openstack: handle safely invalid network name https://review.opendev.org/677501 | 16:07 |
| *** mattw4 has quit IRC | 16:08 | |
| *** sanjayu__ has joined #zuul | 16:10 | |
| openstackgerrit | David Shrewsbury proposed zuul/zuul master: Store autohold requests in zookeeper https://review.opendev.org/661114 | 16:13 |
| *** sanjayu_ has quit IRC | 16:13 | |
| *** jpena is now known as jpena|off | 16:14 | |
| Shrews | tfw you realize your tests fail locally b/c of your running vpn | 16:16 |
| Shrews | :( | 16:16 |
| Shrews | apparently security does not coexist with stability :) | 16:17 |
| fungi | it's an old saw that security and convenience define two ends of a spectrum. maybe with stability in there, three points form a plane | 16:17 |
| fungi | secure, convenient, stable... pick one ;) | 16:17 |
| Shrews | fungi: i pick drinking... is that one? | 16:18 |
| fungi | oh, now you've just defined a three-dimensional field | 16:19 |
| corvus | noorul: can you paste your gate pipeline config? | 16:19 |
| *** noorul has quit IRC | 16:20 | |
| *** noorul has joined #zuul | 16:26 | |
| corvus | noorul: not sure if you saw my last question -- can you paste your gate pipeline config? | 16:27 |
| openstackgerrit | James E. Blair proposed zuul/zuul master: DNM: sample conversion of gerrit to bitbucket job https://review.opendev.org/677510 | 16:29 |
| corvus | noorul, ofosos: ^ there's a mock-up of converting that test to bitbucket; indeed, it does not seem to detect the merge conflict. | 16:30 |
| noorul | corvus: https://etherpad.openstack.org/p/rWL36RmF6W | 16:30 |
| noorul | corvus: from line 112 | 16:30 |
| openstackgerrit | James E. Blair proposed zuul/zuul master: DNM: sample conversion of gerrit to bitbucket job https://review.opendev.org/677510 | 16:31 |
| corvus | noorul, ofosos: ^ actually, that does work, once i switch the gate pipeline from 'independent' to 'dependent'. that was a copy/paste from elsewhere in the bitbucket stack -- i'll go leave a review note for that. | 16:32 |
| corvus | noorul: your pipeline config looks good, and at least a trivial attempt to reproduce this with a unit test isn't showing the error. so i think you may need to dig into what's happening in your actual environment. | 16:33 |
| noorul | Let me raise two more PRs and share the logs | 16:34 |
| *** hwangbo has joined #zuul | 16:35 | |
| *** spsurya has quit IRC | 16:36 | |
| corvus | mordred: take a look at http://zuul.opendev.org/t/openstack/build/99d3f0cd807b4142a359a101ddb5b8ea/console#1/0/7/localhost that comes from http://zuul.opendev.org/t/openstack/build/99d3f0cd807b4142a359a101ddb5b8ea/log/job-output.json#870-885 | 16:44 |
| corvus | mordred: it looks like if we skip an include_role, we don't have any information about what the role was | 16:45 |
| corvus | mordred: do you think we can/should add that to the json output? | 16:45 |
| corvus | clarkb: ^ fyi | 16:45 |
| noorul | corvus: scheduler: http://paste.openstack.org/show/760321/ | 16:48 |
| noorul | corvus: executor: http://paste.openstack.org/show/760322/ | 16:49 |
| openstackgerrit | Jeff Liu proposed zuul/zuul-operator master: Add PerconaXDB Cluster to Zuul-Operator https://review.opendev.org/677315 | 16:49 |
| noorul | ofosos: Are you able to see merge conflict happening in your environment ? | 16:50 |
| corvus | noorul: i see entries for the check pipeline -- we're talking about gate, right? | 16:50 |
| noorul | Yes | 16:50 |
| corvus | noorul: observing the behavior in check won't help identify the problem since check pipelines are independent (zuul has no way of knowing when changes may be approved and therefore conflict in the future). the issue won't show up until you approve both PRs and they are enqueued into gate. | 16:52 |
| noorul | corvus: I think paste is not allowing more than some number of lines | 16:53 |
| corvus | noorul: yes, you might want to paste the scheduler logs starting with when you approved the prs | 16:54 |
| corvus | noorul: there's a length limitation (bytes rather than lines) | 16:54 |
| noorul | corvus: https://www.dropbox.com/s/168t81enxnpt9hm/executor_log_pruned.txt?dl=0 | 16:55 |
| noorul | corvus: https://www.dropbox.com/s/4s0vg9ra36g1cjn/scheduler_log_pruned.txt?dl=0 | 16:55 |
| *** hashar has quit IRC | 16:57 | |
| logan- | having an odd issue running a cross-repo post job http://paste.openstack.org/raw/760325/ "Project X is not allowed to run job Y" despite job Y including project X in its allowed projects. | 17:04 |
| *** jamesmcarthur has quit IRC | 17:04 | |
| fungi | does the job use secrets? | 17:05 |
| logan- | ah yes, there is a parent job a few jobs up the chain of inheritance which uses secrets and does not have allowed-projects updated. i'm sure that's it.. thank you | 17:07 |
| logan- | yep -- works now :) thanks fungi | 17:10 |
| fungi | beware that allowing other projects to use jobs with secrets is risky, at least in cases where the two projects in question are under the control of different parties | 17:11 |
| fungi | (unless those secrets are in trusted config projects, where speculative execution will be disallowed) | 17:11 |
| noorul | Do you know any other users other than those mentioned at https://zuul-ci.org/users.html ? | 17:13 |
| clarkb | noorul: ansible is using it, ofosos said they were going to production soon. Not sure if flaper87 has elasticsearch on it yet but was looking at it. I think armstrongs is also about to go to production? | 17:16 |
| Shrews | easystack does "something" with it | 17:17 |
| noorul | cool | 17:17 |
| fungi | noorul: also https://superuser.openstack.org/articles/zuul-case-study-tungsten-fabric/ | 17:18 |
| fungi | if somebody gets time to summarize that one we can add it to the zuul site too | 17:20 |
| noorul | corvus: I could not find any error in the logs | 17:22 |
| fungi | noorul: oh, also same for https://superuser.openstack.org/articles/continuous-integration-wanted-how-a-large-online-classified-company-uses-zuul/ | 17:22 |
| *** mattw4 has joined #zuul | 17:24 | |
| *** bhavikdbavishi has quit IRC | 17:24 | |
| *** bhavikdbavishi has joined #zuul | 17:27 | |
| ofosos | noorul: im not seeing any merge conflicts. But we're not using the dependent pipeline. I think I know what the problem is. I'll push a fix tomorrow. I'm off for today | 17:27 |
| noorul | ofosos: ok | 17:28 |
| fungi | "tune in tomorrow for the exciting conclusion" | 17:30 |
| fungi | cliffhangers. always leave 'em in suspense | 17:30 |
| ofosos | fungi: sorry, beers | 17:32 |
| fungi | ofosos: you and i share the same sense of priorities! ;) | 17:32 |
| *** jamesmcarthur has joined #zuul | 17:38 | |
| AJaeger | zuul experts, any idea why we see CONFIG_ERROR for periodic requirements jobs? See http://zuul.opendev.org/t/openstack/buildsets?result=CONFIG_ERROR&pipeline=periodic - you might need to look at the log files, I could not get any further information out of this. | 17:41 |
| clarkb | AJaeger: probably related to our chagne to add ref info to the subject of the smtp reporter | 17:43 |
| clarkb | I'm guessing change.ref isn't set after all? | 17:43 |
| clarkb | I guess we should revert that change | 17:44 |
| *** sshnaidm is now known as sshnaidm|afk | 17:44 | |
| fungi | or add something we know will be present and conditionally use one or the other (zuul.branch maybe when zuul.ref is missing?) | 17:45 |
| clarkb | fungi: I think zuul.ref is the one that is always set then you get additional info like branch or tag if the type of ref can be determined | 17:45 |
| clarkb | I think the problem with periodic jobs is that they are not triggered by an action on a ref, they are just run by a timer | 17:46 |
| *** sanjayu__ has quit IRC | 17:46 | |
| corvus | clarkb: the timer enqueues a model.Ref for each branch head | 17:46 |
| corvus | actually, it should enqueue a model.Branch, which is descended from model.Ref, but it's the ref object that supplies the attrs for formatting | 17:47 |
| *** igordc has joined #zuul | 17:48 | |
| openstackgerrit | Tristan Cacqueray proposed zuul/zuul-jobs master: doc: note that mirror_info isn't actually implemented https://review.opendev.org/677529 | 18:04 |
| tristanC | not sure about https://review.opendev.org/677529, but i couldn't find any roles that uses mirror_info | 18:04 |
| AJaeger | speaking about mirror_info, there's also https://review.opendev.org/675703 for review - corvus, did you see that one? | 18:05 |
| AJaeger | tristanC: Yes, it's not implemented yet - still in the design phase | 18:05 |
| *** jamesmcarthur has quit IRC | 18:05 | |
| corvus | well, it sort of went back to the design phase :) | 18:06 |
| *** jamesmcarthur has joined #zuul | 18:06 | |
| corvus | but yeah, we can merge tristanC's change until we have the first role using it | 18:06 |
| tristanC | corvus: or perhaps the mirror.rst shouldn't be included in the published doc? | 18:07 |
| corvus | that's really hard to do with our jobs | 18:08 |
| corvus | easiest way to make it correct is to get to work implementing it :) i'll start on the operating system role after lunch | 18:11 |
| AJaeger | corvus, tristanC, do we need https://review.opendev.org/676836 for swift logging? Or is it wrong to do as questions indiciate? "base: generate the zuul-manifest before ara-report". Should we test this change via base-test first? | 18:11 |
| *** jamesmcarthur has quit IRC | 18:12 | |
| *** jamesmcarthur has joined #zuul | 18:13 | |
| corvus | AJaeger: best way to answer the question for sure is base-test | 18:13 |
| tristanC | AJaeger: i'll WIP that, it seems like ara-report could generate an artifact so that it's accessible from the summary | 18:14 |
| AJaeger | tristanC: let me -1 - if we want this, we need it in base-test first... | 18:17 |
| AJaeger | thanks, tristanC and corvus | 18:17 |
| corvus | AJaeger, tristanC: maybe create an empty ara-report directory before running the manifest and ara-generate roles | 18:21 |
| corvus | tristanC, AJaeger: another idea to think about -- have the manifest role take a list of directories not to recurse into | 18:22 |
| corvus | tristanC, AJaeger: yet another idea -- have the web app ignore ara-report (do it in actions) | 18:23 |
| * corvus stops brainstorming and afks | 18:23 | |
| openstackgerrit | Merged zuul/zuul-jobs master: Update documentation for Debian/Ubuntu mirror_info https://review.opendev.org/675703 | 18:25 |
| *** bjackman has joined #zuul | 18:29 | |
| *** noorul has quit IRC | 18:32 | |
| *** electrofelix has quit IRC | 18:35 | |
| openstackgerrit | Merged zuul/zuul-jobs master: doc: note that mirror_info isn't actually implemented https://review.opendev.org/677529 | 18:45 |
| fungi | following up on the earlier leboncoin case study earlier, someone just pointed this comment out to me: https://twitter.com/gchenuet/status/1163860245443207168 | 18:55 |
| fungi | seems they're still fans ;) | 18:56 |
| Shrews | corvus: is this a known random failure? https://ae2197746e23c7b714c7-0b8936c820696cdb6c8467dc1e642b32.ssl.cf1.rackcdn.com/661114/13/check/tox-py36/4a2da8a/testr_results.html.gz | 18:56 |
| fungi | takes a lot of scrolling to get to that traceback | 18:57 |
| corvus | Shrews: news to me | 18:58 |
| fungi | i haven't seen that one before that i can recall | 18:58 |
| Shrews | tox-py35 passed | 18:58 |
| *** jamesmcarthur has quit IRC | 18:58 | |
| corvus | 15 is a lot of builds, so it's probably right on the edge | 18:59 |
| tristanC | corvus: it seems like "having the manifest role takes a list of directories not to recurse into" is the most efficient improvement | 19:10 |
| tristanC | using ["ara-report"] as a default value | 19:10 |
| *** bjackman has quit IRC | 19:14 | |
| *** jamesmcarthur has joined #zuul | 19:28 | |
| openstackgerrit | Tristan Cacqueray proposed zuul/zuul-jobs master: generate-zuul-manifest: add excludes list https://review.opendev.org/677556 | 19:32 |
| *** jamesmcarthur has quit IRC | 19:59 | |
| *** jamesmcarthur has joined #zuul | 20:00 | |
| openstackgerrit | Tristan Cacqueray proposed zuul/zuul-jobs master: generate-zuul-manifest: add excludes list https://review.opendev.org/677556 | 20:07 |
| corvus | fungi: what was the reason we didn't make 'type' a list in 675703 ? | 20:34 |
| *** jamesmcarthur has quit IRC | 20:40 | |
| *** jamesmcarthur_ has joined #zuul | 20:40 | |
| fungi | good question, i think we certainly could. once i get dinner out of the way i'll double-check that there wasn't a good reason | 20:41 |
| fungi | though frankly, the odds it will be anything other than "deb" are slim since "deb-src" would only be if you wanted to be able to fetch source packages or install build dependencies of source packages | 20:42 |
| corvus | yeah, but we're aiming for future-proofing here :) | 20:43 |
| corvus | *someone* might want that :) | 20:43 |
| *** jamesmcarthur_ has quit IRC | 20:47 | |
| *** bhavikdbavishi has quit IRC | 20:47 | |
| fungi | the debian project might even want to do that ;) | 20:47 |
| fungi | but yeah, i think we ought to be able to switch it for improved consistency | 20:47 |
| fungi | if you need to not have matrix expansion for some set, just make a separate entry | 20:47 |
| fungi | i'll whip up a followup change shortly | 20:48 |
| corvus | fungi: don't worry about it -- i'm working on an implementation, i'll include it there | 20:48 |
| fungi | wfm, thanks! | 20:49 |
| *** jamesmcarthur has joined #zuul | 20:55 | |
| *** jamesmcarthur has quit IRC | 20:58 | |
| *** jamesmcarthur has joined #zuul | 20:58 | |
| *** jamesmcarthur has quit IRC | 21:04 | |
| openstackgerrit | Tristan Cacqueray proposed zuul/zuul-jobs master: generate-zuul-manifest: add excludes list https://review.opendev.org/677556 | 21:11 |
| corvus | fungi: i think we may also need to add something to filter for architectures | 21:17 |
| corvus | fungi: with the current role, on aarch64, we add '-ports' to the url | 21:17 |
| corvus | hrm... i guess we could do the same here and make it automatic | 21:18 |
| corvus | or we wanted it to be explicit, we could add it to the mirror_info variable... | 21:18 |
| fungi | yeah, at least for debian the ports tree is independent and could be on different servers even | 21:20 |
| fungi | i don't know that i would break up the url at all, easier to treat it as a single string | 21:21 |
| fungi | i'd rather not make assumptions about how someone's mirror is organized | 21:22 |
| corvus | fungi: so are you voting for adding "architectures: ['amd64', 'aarch65']" to the list? | 21:22 |
| fungi | er | 21:23 |
| fungi | i'm slightly confused by the example | 21:23 |
| corvus | modulo typos :) | 21:23 |
| fungi | ['amd64', 'arm64'] i guess... let me check | 21:24 |
| corvus | fungi: it would be a list of architectures hosted at the url specified in the list, then the role would include that list item iff the current architecture matched one in the list | 21:24 |
| fungi | normally you wouldn't do cross/multi-arch sources lists | 21:24 |
| fungi | oh, i see | 21:24 |
| corvus | if it's not possible (or unfeasible) we could make that a singleton rather than list | 21:25 |
| fungi | because you'll have systems for each architecture which need different urls | 21:25 |
| corvus | the existing values of ansible_architecture which are relevant seem to be 'x86_64' and 'aarch64' | 21:26 |
| fungi | got it, which will somewhere be translated to amd64 and arm64 | 21:26 |
| fungi | or not translated at all, simply used by ansible for selecting? | 21:27 |
| corvus | i don't think we have to write the arch into the config file, so i think it's really 2 separate things | 21:27 |
| corvus | yeah that | 21:27 |
| corvus | ansible_architecture for selecting, then apt handles the rest | 21:27 |
| fungi | so yeah, maybe it needs to be the ansible architectures, not the debian architectures | 21:27 |
| fungi | where it likely gets hairy is if you want to cross-build and install packages for one architecture on a machine of a different architecture, then you'll need to translate from debian architecture names to ansible architecture names | 21:29 |
| fungi | but i'm happy to consider that a corner case | 21:29 |
| fungi | (the amd64/arm64 example is not really a great one, since both are official architectures for debian now) | 21:30 |
| corvus | i guess you could say "include the arm64 sources entries on this amd64 host" that should work right? | 21:31 |
| fungi | in opendev's case the url for both would be identical | 21:31 |
| fungi | but yeah | 21:31 |
| corvus | really the only reason not to include them all is just so we're not bulking up the package list on our x86 hosts | 21:31 |
| fungi | to include the arm64 sources entries you'd need to know to look up the aarch64 set | 21:31 |
| fungi | not a big deal | 21:31 |
| corvus | right, but with the system we're talking about, the admin could say "architectures: ['x86_64']" for a url which hosts arm packages and get the full set | 21:32 |
| fungi | well, every additional entry is another (set of) http gets and possibly gpg signature validations and index additions and so on... so increases the time/workload unnecessarily | 21:33 |
| fungi | (as the reason not to include them all always) | 21:33 |
| corvus | yeah, that seems to make this worth doing | 21:33 |
| fungi | agreed | 21:33 |
| corvus | okay, i'll do it like that, thanks :) | 21:34 |
| fungi | thank *you*! | 21:34 |
| openstackgerrit | Tristan Cacqueray proposed zuul/zuul-jobs master: generate-zuul-manifest: add excludes list https://review.opendev.org/677556 | 21:36 |
| *** mattw4 has quit IRC | 21:43 | |
| *** mattw4 has joined #zuul | 21:51 | |
| *** mattw4 has quit IRC | 22:02 | |
| *** mattw4 has joined #zuul | 22:07 | |
| *** zbr has quit IRC | 22:14 | |
| *** zbr has joined #zuul | 22:15 | |
| *** armstrongs has joined #zuul | 22:18 | |
| *** armstrongs has quit IRC | 22:29 | |
| openstackgerrit | James E. Blair proposed zuul/zuul-jobs master: Add configure-os-mirrors role https://review.opendev.org/677578 | 22:37 |
| corvus | fungi, tristanC, clarkb, AJaeger: ^ | 22:37 |
| corvus | i'm hoping the rest will be much simpler than that | 22:38 |
| *** jeliu_ has quit IRC | 22:40 | |
| clarkb | corvus: left some comments on that, but it seems to be fairly compelte for the distro pacakge repo mirrors | 22:54 |
| *** mattw4 has quit IRC | 23:03 | |
| *** zbr has quit IRC | 23:22 | |
| *** zbr has joined #zuul | 23:23 | |
| openstackgerrit | Ian Wienand proposed zuul/zuul-jobs master: upload-logs-swift: fix keystoneauth1 exceptions https://review.opendev.org/677584 | 23:57 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!