| *** ssbarnea|rover has quit IRC | 00:13 | |
| *** ssbarnea|rover has joined #zuul | 00:17 | |
| *** sdake has quit IRC | 00:19 | |
| *** saneax has quit IRC | 00:26 | |
| *** sdake has joined #zuul | 00:28 | |
| *** rlandy has quit IRC | 00:51 | |
| sean-k-mooney | pabelanger: corvus: thanks for your help i tested https://review.openstack.org/#/c/633796/1 with my third party ci and it now works https://review.seanmooney.info/c/test/+/1041#message-9f4b6a8a00de3de14197c8a48466df27b1412748 | 01:14 |
|---|---|---|
| sean-k-mooney | pabelanger: i have also optimised my zuul config as you suggested to use a singel gerrit connection for all openstack related repos | 01:15 |
| *** sdake has quit IRC | 01:19 | |
| *** sdake has joined #zuul | 01:23 | |
| pabelanger | sean-k-mooney: cool! | 02:36 |
| *** sdake has quit IRC | 03:00 | |
| *** bhavikdbavishi has joined #zuul | 03:09 | |
| *** bhavikdbavishi has quit IRC | 03:38 | |
| *** bhavikdbavishi has joined #zuul | 03:44 | |
| *** chkumar|out is now known as chandankumar | 04:38 | |
| *** bhavikdbavishi has quit IRC | 05:37 | |
| *** bhavikdbavishi has joined #zuul | 05:44 | |
| *** bhavikdbavishi has quit IRC | 05:57 | |
| *** bhavikdbavishi has joined #zuul | 06:17 | |
| *** badboy has joined #zuul | 06:23 | |
| *** swest has joined #zuul | 06:30 | |
| *** quiquell|off is now known as quiquell | 06:39 | |
| *** jesusaur has quit IRC | 07:08 | |
| *** jesusaur has joined #zuul | 07:12 | |
| *** saneax has joined #zuul | 07:18 | |
| *** bhavikdbavishi has quit IRC | 08:03 | |
| *** gtema has joined #zuul | 08:10 | |
| *** pcaruana has joined #zuul | 08:11 | |
| *** avass has joined #zuul | 08:11 | |
| *** gtema has quit IRC | 08:14 | |
| *** gtema has joined #zuul | 08:18 | |
| openstackgerrit | Fabien Boucher proposed openstack-infra/nodepool master: Add ip-pool option to the openstack provider https://review.openstack.org/619525 | 08:19 |
| openstackgerrit | Tristan Cacqueray proposed openstack-infra/zuul master: web: add /{tenant}/buildsets route https://review.openstack.org/630035 | 08:19 |
| *** themroc has joined #zuul | 08:21 | |
| *** gtema has quit IRC | 08:24 | |
| *** gtema has joined #zuul | 08:38 | |
| *** bhavikdbavishi has joined #zuul | 08:41 | |
| quiquell | Hello | 08:44 |
| quiquell | Is there any way to share an openstack tenant with two nodepool-launchers ? | 08:44 |
| quiquell | So they don't clean each other images I suppose it's an issue with quotas and the like | 08:44 |
| *** electrofelix has joined #zuul | 08:51 | |
| *** jpena|off is now known as jpena | 08:57 | |
| *** saneax has quit IRC | 09:25 | |
| *** saneax has joined #zuul | 09:27 | |
| badboy | can I use an ssh command on a nodepool worker node? | 10:04 |
| badboy | I would like to add a comment as a part of a playbook | 10:05 |
| tobiash | quiquell: I think if you name the providers of the config differently it might work | 10:12 |
| quiquell | sshnaidm: ^ | 10:12 |
| tobiash | quiquell: but no guarantee about that ;) | 10:13 |
| tobiash | and quota management won't work | 10:13 |
| quiquell | tobiash: we don't care about that I think | 10:14 |
| quiquell | tobiash: ack will look thanks ! | 10:14 |
| tobiash | badboy: what do you mean exactly? | 10:14 |
| tobiash | badboy: nodepool only checks ssh availability and retrieves host keys, it doesn't do a login or scripting on nodes | 10:15 |
| badboy | tobiash: I was talking about Zuul | 10:15 |
| tobiash | badboy: ansible by default uses ssh, what's your use case? | 10:16 |
| tobiash | badboy: ah, you want to run ssh from the worker node to somewhere else? | 10:16 |
| tobiash | badboy: that's easy with a shell task | 10:17 |
| badboy | tobiash: yes | 10:17 |
| tobiash | you might need to use secrets to add a private ssh key before though | 10:17 |
| badboy | tobiash: I want to do something like this: ssh -p 29418 my-gerrit.example.com gerrit review -m $'"Some text"' <change_id> | 10:18 |
| badboy | tobiash: the same user exists on the Zuul server and the worker node | 10:18 |
| tobiash | badboy: yes, you can simply run this in a shell task if you prepared the private key before that | 10:18 |
| badboy | tobiash: with the same key | 10:18 |
| *** sshnaidm has quit IRC | 10:19 | |
| tobiash | badboy: you have two choices then: either add that key as a secret or mount that key into the job workspace using (un)trusted_rw_paths config of the executor | 10:20 |
| tobiash | I'd suggest the secret, because (un)trusted_rw/ro_paths are mounted into *every* job which is probably not what you want | 10:20 |
| badboy | tobiash: the key is already in the ~/.ssh/id_rsa | 10:20 |
| tobiash | on the node? | 10:20 |
| badboy | yes | 10:21 |
| tobiash | ok, then just run ssh in a shell task | 10:21 |
| tobiash | then it's really easy | 10:21 |
| tobiash | (but also unsafe) ;) | 10:21 |
| badboy | tobiash: I just want to get it all started | 10:23 |
| badboy | tobiash: after that I will fix that | 10:23 |
| badboy | tobiash: I keep getting Host key verification failed | 10:24 |
| badboy | tobiash: on that worker-node | 10:25 |
| badboy | tobiash: even though it's using the correct user and the keys are ok | 10:25 |
| tobiash | badboy: host key verification is not about authentication | 10:25 |
| tobiash | you need to either gather the host keys and put them into known_hosts or just ignore host key verification | 10:26 |
| *** sshnaidm has joined #zuul | 10:26 | |
| tobiash | badboy: you can disable it in the ssh command line: ssh -o StrictHostKeyChecking=no | 10:27 |
| quiquell | tobiash: Has found a real weird stuff | 10:28 |
| quiquell | tobiash: looks like having a jinja stuff at git comment breaks zuul | 10:28 |
| quiquell | tobiash: https://review.rdoproject.org/r/#/c/18623/ | 10:29 |
| badboy | tobiash: thx, adding to known_hosts worked | 10:29 |
| badboy | tobiash: one more, which zuul variable contains the commit sha? | 10:30 |
| tobiash | check out the inventory, it contains all variables | 10:35 |
| tobiash | badboy: the vars are also documented here: https://zuul-ci.org/docs/zuul/user/jobs.html#zuul-variables | 10:39 |
| badboy | tobiash: I am reading that atm and still cannot extract the change Id or commit id | 10:40 |
| tobiash | badboy: change id and patchset are its own variables | 10:40 |
| tobiash | here: https://zuul-ci.org/docs/zuul/user/jobs.html#var-zuul.items.change | 10:40 |
| tobiash | badboy: but if you want zuul to report line based comments, check this out: https://zuul-ci.org/docs/zuul/user/jobs.html#leaving-file-comments | 10:41 |
| tobiash | in case that fits your needs, zuul itself will take care about reporting line based comments | 10:41 |
| badboy | tobiash: so I should use {{ zuul.items[].change }} instead of {{ zuul.change }} ? | 10:42 |
| tobiash | badboy: oh, you can use zuul.change and zuul.patchset | 10:44 |
| tobiash | looks like these two are undocumented | 10:44 |
| badboy | tobiash: https://zuul-ci.org/docs/zuul/user/jobs.html#var-zuul.change | 10:44 |
| tobiash | oh, overlooked that | 10:45 |
| badboy | tobiash: http://paste.openstack.org/show/744244/ | 10:46 |
| badboy | tobiash: I would like to extract the Change-Id | 10:47 |
| badboy | tobiash: currently it's stored as a part of the commit message | 10:47 |
| tobiash | that's a gerrit thing, so you'll need some jinja/bash magic to extract it | 10:48 |
| tobiash | but typically the gerrit api works with the change number and patchset | 10:48 |
| tobiash | so if your use case is to comment, you don't need the Change-Id from the commit message | 10:49 |
| badboy | tobiash: ssh -p 29418 my-gerrit.example.com gerrit review -m $'"Some text" {{ zuul.change }},{{ zuul.patchset }} | 10:54 |
| badboy | tobiash: ^ that's working except for the newline chars | 10:54 |
| tobiash | cool | 10:54 |
| badboy | tobiash: thanks for your help | 10:56 |
| tobiash | yw | 10:58 |
| *** bhavikdbavishi has quit IRC | 11:02 | |
| *** avass has quit IRC | 11:03 | |
| *** gtema has quit IRC | 11:20 | |
| electrofelix | hughsaunders: are you still working on the nodepool agent plugin for Jenkins? | 11:31 |
| *** gtema has joined #zuul | 11:48 | |
| tristanC | quiquell: that's indeed a weird stuff, turns out the commit message ends up not escaped in the inventory and jinja expression are evaluated by template task :-) | 11:53 |
| quiquell | tristanC: weird... | 11:54 |
| quiquell | tristanC: crearly a bug, going to put one at story board | 11:54 |
| quiquell | will try to debug | 11:54 |
| *** luizbag has joined #zuul | 11:58 | |
| quiquell | tristanC: I see we have the commit message in the invetory, is this new ? | 11:59 |
| tristanC | quiquell: you can debug the zuul.message variable, and for some reason it get "jinja" evaluated in any task using template | 11:59 |
| quiquell | tristanC: going to throw a review for this | 11:59 |
| quiquell | tristanC: I suppose inventory.yaml itself is a jinja template so weird stuff hapend like nested expansion | 12:00 |
| *** sdake has joined #zuul | 12:02 | |
| tristanC | then we might want to escape jinja expressions before storing the message in the inventory... | 12:02 |
| quiquell | tristanC: Something like that I am looking at the code | 12:02 |
| *** jpena is now known as jpena|lunch | 12:04 | |
| *** bhavikdbavishi has joined #zuul | 12:05 | |
| tristanC | quiquell: it should be done somewhere like https://git.zuul-ci.org/cgit/zuul/tree/zuul/executor/client.py#n181 | 12:09 |
| quiquell | tristanC: Yep I am reproducing it first with a unit test | 12:09 |
| quiquell | Then fix | 12:09 |
| *** steddy has joined #zuul | 12:16 | |
| openstackgerrit | Quique Llorente proposed openstack-infra/zuul master: Escape jinja2 stuff from inventory https://review.openstack.org/633930 | 12:30 |
| steddy | Hi, I'm trying to debug a zuul installation. I followed the 'Zuul from scratch' installation guide and have connected my install to github. I'm trying to debug an issue where I recheck a pull request, then pipeline briefly flashes up in the status page but then dissappears with no logging. Can anyone help me get some debug logs to troubleshoot the issue. Thanks | 12:36 |
| *** sdake has quit IRC | 12:43 | |
| *** quiquell is now known as quiquell|lunch | 12:45 | |
| *** sdake has joined #zuul | 12:45 | |
| tobiash | quiquell: does it fail to completely run the job containing jinja stuff in the commit message or does it just do weird stuff if you want to do something with it? | 12:46 |
| tobiash | ah, looked at that link and everything is in retry limit | 12:47 |
| *** rlandy has joined #zuul | 12:54 | |
| *** quiquell|lunch is now known as quiquell | 12:57 | |
| tristanC | tobiash: it seems like zuul variables are evaluated for any task using template | 12:57 |
| tobiash | k | 12:58 |
| *** jpena|lunch is now known as jpena | 13:02 | |
| openstackgerrit | Tobias Henkel proposed openstack-infra/zuul master: Make git clone timeout configurable https://review.openstack.org/633936 | 13:08 |
| *** badboy has quit IRC | 13:14 | |
| *** gtema has quit IRC | 13:25 | |
| dmsimard | FYI: ara 0.16.3 was just tagged and released to address a regression in the upcoming release of Ansible 2.8 | 13:31 |
| *** bhavikdbavishi has quit IRC | 13:33 | |
| openstackgerrit | Benedikt Löffler proposed openstack-infra/zuul master: Report the build execute_time via mqtt https://review.openstack.org/633945 | 13:47 |
| *** pcaruana has quit IRC | 13:50 | |
| openstackgerrit | Sorin Sbarnea proposed openstack-infra/zuul-jobs master: Fix failure to add user to docker group on centos https://review.openstack.org/633948 | 13:54 |
| *** pcaruana has joined #zuul | 13:57 | |
| openstackgerrit | David Shrewsbury proposed openstack-infra/nodepool master: WIP Revert "Revert "Add a timeout for the image build"" https://review.openstack.org/633792 | 13:59 |
| quiquell | tobiash, tristanC: Maybe we can escape jinja stuff at https://github.com/openstack-infra/zuul/blob/master/zuul/executor/server.py#L588 ? | 14:06 |
| tobiash | quiquell: yes, I think that's the most generic place where you could do it | 14:08 |
| quiquell | tobiash: Will just do it for zuul.message | 14:08 |
| tobiash | quiquell: we should think about doing this for all values in that dict | 14:09 |
| quiquell | tobiash: this can have performance penalty | 14:09 |
| tobiash | because any jinja expression in that dict will brean this | 14:09 |
| tobiash | quiquell: how long do you expect this to take? | 14:09 |
| quiquell | tobiash: only for "zuul" herarchy ? | 14:09 |
| quiquell | tobiash: I see a lot of stuff here http://logs.rdoproject.org/21/18621/8/check/tripleo-ci-reproducer-fedora-28-libvirt/7c25ec9/zuul-info/inventory.yaml | 14:10 |
| tobiash | I guess escaping the zuul hierarchy will just take a few milliseconds | 14:10 |
| quiquell | tobiash: most of it is mechanical | 14:10 |
| quiquell | tobiash: the probability that it has jinja is low, but commit message having jinaj is high | 14:10 |
| tobiash | well, escaping is just a few string operations right? | 14:11 |
| tobiash | so solving it in the whole zuul hierarchy prevents us from future occurrences of jinja stuff | 14:11 |
| quiquell | tobiash: let's escape all is more secure we can improve later on | 14:11 |
| tobiash | I'm pretty sure you won't be able to noteice any performance difference... | 14:12 |
| tobiash | the git fetch and merge is probably thousands of times slower than the escaping ;) | 14:12 |
| quiquell | tobiash: what is the best unit test to test this ? | 14:13 |
| quiquell | tobiash: have try with the wrong one | 14:13 |
| tobiash | quiquell: I think you started good in test_inventory | 14:14 |
| quiquell | tobiash: ack will continue there | 14:14 |
| tobiash | of course you need to change the assert to expect the escaped value | 14:14 |
| quiquell | tobiash: yep was just playing around to understand that | 14:15 |
| quiquell | tobiash: going to scape {{ with {{ '{{' }} and the counter part | 14:18 |
| quiquell | is this good ? | 14:18 |
| tobiash | I think that will probably work | 14:18 |
| *** sdake has quit IRC | 14:20 | |
| adam_g | anyone using the github driver in production? | 14:27 |
| pabelanger | yes | 14:28 |
| pabelanger | ansible-network / awx / ansible-runner are gatting with zuulv3 and github | 14:28 |
| pabelanger | tobiash is also using GHE | 14:28 |
| adam_g | pabelanger: does it get very much traffic? | 14:28 |
| pabelanger | adam_g: not as much as tobiash, but a bit | 14:29 |
| tobiash | adam_g: we're using it to gate ~1000 repos | 14:29 |
| tobiash | (on ghe) | 14:30 |
| tobiash | and it didn't kill our github in the last year ;) | 14:31 |
| adam_g | ah | 14:31 |
| adam_g | any stats on how many events per hour its typically handling? | 14:31 |
| adam_g | we're running one against an internal GHE and find the queue grinds to a halt with any significant number of incoming events | 14:32 |
| adam_g | granted its an older codebase (2.5 + the experimental github integration patches), but looking at master the issues im seeing should still be there | 14:32 |
| tobiash | I'd first have to request stats of our github | 14:32 |
| adam_g | we started getting rate limited hard by our GHE, but its because they put a frontend LB in front that broke etag caching | 14:33 |
| tobiash | but you should at least upgrade to ghe 2.12 which introduces github app auth (which I'd recommend) | 14:33 |
| Shrews | oh lookie, adam_g does still exist :) | 14:33 |
| adam_g | tobiash: interesting, ill check into that. we probably support that but dont use it | 14:34 |
| adam_g | i did find getChange() for GH is god awful slow, and is the source of our problem | 14:34 |
| tobiash | adam_g: when using github app auth, each installation has its own rate limit btw | 14:34 |
| tobiash | so when running a large zuul I'd strongly recommend gh app auth | 14:34 |
| *** sdake has joined #zuul | 14:34 | |
| pabelanger | +1 | 14:35 |
| adam_g | tobiash: whats the app auth buy wrt network round trips? | 14:35 |
| adam_g | i found shortcutting github3 a little bit sped things up quite a bit | 14:35 |
| adam_g | from ~48sec for 24 changes to 11sec | 14:36 |
| adam_g | Shrews: o/ | 14:36 |
| tobiash | adam_g: app auth doesn't change the number or kind of api calls | 14:37 |
| adam_g | event queue processing takes many minutes while the run_handler lock is held and the wohle scheduler grinds to a halt trying to catch up | 14:37 |
| *** bhavikdbavishi has joined #zuul | 14:38 | |
| adam_g | i can imagine it gets much worse when getChange() is called recursively for Depends-on | 14:39 |
| tobiash | well, I already had a year of optimization of the github driver ;) | 14:40 |
| tobiash | the state of one year ago probably would kill our github now | 14:41 |
| adam_g | yea.. ive looked at git history trying to find key optimizations. the problem isnt on the github side for us. enabling debug logging for requests/github3.py provided some alarming insight :) | 14:43 |
| tobiash | adam_g: btw, zuul hooks a caching layer into the requests session of github3.py | 14:44 |
| adam_g | yep | 14:45 |
| adam_g | it still requires roundtrips for the conditional GET | 14:45 |
| adam_g | what ive seen is it churns for a loooong time gathering user permissions | 14:46 |
| tobiash | that's true | 14:46 |
| adam_g | e.g., https://github.com/openstack-infra/zuul/blob/master/zuul/driver/github/githubconnection.py#L1283 | 14:47 |
| *** quiquell is now known as quiquell|off | 14:48 | |
| adam_g | i wish the people maintaining our zuul could get it upgraded to 3.0/master | 14:48 |
| adam_g | but i will probably try to forward-port some of these patches in the meantime | 14:49 |
| tobiash | adam_g: you're on zuul2? | 14:49 |
| *** sdake has quit IRC | 14:49 | |
| adam_g | tobiash: yep. :( | 14:50 |
| adam_g | http://paste.openstack.org/show/744254/ | 14:53 |
| adam_g | i havent tested but looking thru master it seems like those numbers would be similar | 14:54 |
| *** sdake has joined #zuul | 14:54 | |
| adam_g | was mostly curious to know if the scheduler changed in some way where that slowness wouldnt effect other queue processing | 14:55 |
| tobiash | adam_g: most github api requests are done in the driver loop atm, so the queue processing is mostly unaffected | 14:55 |
| openstackgerrit | Quique Llorente proposed openstack-infra/zuul master: Escape jinja2 stuff from inventory https://review.openstack.org/633930 | 14:56 |
| tobiash | at least all api requests that do get change etc from events | 14:56 |
| adam_g | tobiash: oh interesting, which is the driver loop? | 14:59 |
| *** themroc has quit IRC | 15:00 | |
| tobiash | adam_g: https://git.zuul-ci.org/cgit/zuul/tree/zuul/driver/github/githubconnection.py#n427 | 15:04 |
| adam_g | tobiash: oh, nice | 15:04 |
| tobiash | that preprocesses the webhooks and inserts trigger events into the scheduler main loop | 15:05 |
| adam_g | cool. more ammo to throw at our CICD people to upgrade :) | 15:07 |
| pabelanger | tobiash: corvus: clarkb: I wouldn't mind if we discussed maybe a zuul release this week, to help pick up recent github fix for issue we had in SF.io, along with gearman client keepalive. | 15:09 |
| openstackgerrit | Benedikt Löffler proposed openstack-infra/zuul master: Fix flake8 error: E117 over-indented https://review.openstack.org/633974 | 15:10 |
| *** sdake has quit IRC | 15:11 | |
| *** sdake has joined #zuul | 15:14 | |
| *** quiquell|off has quit IRC | 15:19 | |
| openstackgerrit | David Shrewsbury proposed openstack-infra/nodepool master: Revert "Revert "Add a timeout for the image build"" https://review.openstack.org/633792 | 15:25 |
| Shrews | corvus: that ^ version seems to work as a daemon (i was getting log file output, at least) | 15:26 |
| *** quiquell has joined #zuul | 15:33 | |
| *** quiquell is now known as quiquell|off | 15:33 | |
| Shrews | i wonder if we should try it first on nb03. we don't seem to have a real good way to test the builder as a daemon :/ | 15:40 |
| tobiash | pabelanger: sounds good to me, are there any known issues with the new containers? | 15:41 |
| pabelanger | tobiash: I am unsure, I haven't been following along much on the container work | 15:43 |
| *** arxcruz|ruck is now known as arxcruz | 15:58 | |
| *** pcaruana has quit IRC | 16:01 | |
| *** gtema has joined #zuul | 16:01 | |
| *** saneax has quit IRC | 16:01 | |
| *** luizbag has quit IRC | 16:03 | |
| *** steddy has quit IRC | 16:03 | |
| openstackgerrit | Tobias Henkel proposed openstack-infra/zuul master: github: log the project name when getBranches fails https://review.openstack.org/583841 | 16:10 |
| *** luizbag has joined #zuul | 16:10 | |
| tobiash | tristanC: I've picked up this, I hope that's ok for you ^ | 16:10 |
| openstackgerrit | Tobias Henkel proposed openstack-infra/zuul master: github: log the project name when getBranches fails https://review.openstack.org/583841 | 16:13 |
| *** sdake has quit IRC | 16:14 | |
| *** pcaruana has joined #zuul | 16:17 | |
| *** ssbarnea|rover has quit IRC | 16:20 | |
| *** ssbarnea has joined #zuul | 16:21 | |
| *** sdake has joined #zuul | 16:22 | |
| openstackgerrit | Merged openstack-infra/zuul master: Fix flake8 error: E117 over-indented https://review.openstack.org/633974 | 16:39 |
| *** pcaruana has quit IRC | 16:45 | |
| *** electrofelix has quit IRC | 16:49 | |
| *** sdake has quit IRC | 16:51 | |
| *** sdake has joined #zuul | 17:01 | |
| openstackgerrit | Sorin Sbarnea proposed openstack-infra/zuul-jobs master: Fix failure to add user to docker group on centos https://review.openstack.org/633948 | 17:01 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: WIP add provides/requires support https://review.openstack.org/633605 | 17:07 |
| *** bhavikdbavishi has quit IRC | 17:19 | |
| *** bhavikdbavishi has joined #zuul | 17:19 | |
| *** gtema has quit IRC | 17:37 | |
| *** bhavikdbavishi has quit IRC | 17:58 | |
| *** saneax has joined #zuul | 18:10 | |
| *** jpena is now known as jpena|off | 18:13 | |
| *** pcaruana has joined #zuul | 18:17 | |
| *** bhavikdbavishi has joined #zuul | 18:19 | |
| *** themroc has joined #zuul | 18:23 | |
| *** saneax has quit IRC | 18:23 | |
| *** sshnaidm is now known as sshnaidm|afk | 18:25 | |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Add provides/requires support https://review.openstack.org/633605 | 18:29 |
| adam_g | tobiash: so i was just looking closer at the driver loop you pointed me at. i dont think that really solves the issue ive been hitting. what im seeing is processing of the trigger events in the scheduler's trigger_event_queue taking so long that on a busy server it never manages to work thru the queue and never leaving here https://github.com/openstack-infra/zuul/blob/master/zuul/scheduler.py#L1009 | 18:38 |
| SpamapS | adam_g: moar CPU | 18:46 |
| adam_g | SpamapS: no, its *lots* of network round trips to GH for every change | 18:47 |
| SpamapS | adam_g: ah, I thought the GH bits happened before that loop. | 18:48 |
| SpamapS | adam_g: honestly, etag breakage sounds like your problem, not Zuul. | 18:48 |
| SpamapS | adam_g: but also the app model would probably help a lot since it might give you more rate limit pools | 18:49 |
| adam_g | SpamapS: its getChange() in the connection that does a buncha stuff. | 18:49 |
| adam_g | SpamapS: nah etags are actually working again as expected but still 1 require API round trip per for a hit, and double for miss :) | 18:50 |
| adam_g | *require 1 API RT | 18:50 |
| adam_g | the one that really sucks is this https://github.com/openstack-infra/zuul/blob/master/zuul/driver/github/githubconnection.py#L1149 | 18:51 |
| adam_g | for really active PRs especially | 18:51 |
| SpamapS | adam_g: I wonder if you could spin up a v3 and let it consume all the hooks for a while, see if it does better (maybe by multiplexing the hooks) | 18:51 |
| SpamapS | also I wonder if you could actually benefit from a thread pool. | 18:52 |
| SpamapS | oh and yeah, that particular method caused us concern back in the Bonny days, I remember. | 18:53 |
| adam_g | SpamapS: id love to but dont have any cycles to do it, im already punting on other work to debug this since nobody else around here seems able to :) | 18:53 |
| adam_g | yep | 18:53 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Add provides/requires support https://review.openstack.org/633605 | 19:12 |
| *** luizbag has quit IRC | 19:17 | |
| *** panda is now known as panda|off | 19:59 | |
| openstackgerrit | David Shrewsbury proposed openstack-infra/nodepool master: Revert "Revert "Add a timeout for the image build"" https://review.openstack.org/633792 | 20:01 |
| *** sdake has quit IRC | 20:06 | |
| *** sdake has joined #zuul | 20:06 | |
| *** bhavikdbavishi has quit IRC | 20:10 | |
| *** rfolco has quit IRC | 20:13 | |
| openstackgerrit | David Shrewsbury proposed openstack-infra/nodepool master: Revert "Revert "Add a timeout for the image build"" https://review.openstack.org/633792 | 20:20 |
| *** rfolco has joined #zuul | 20:24 | |
| *** sdake has quit IRC | 20:39 | |
| *** rfolco has quit IRC | 20:40 | |
| *** sdake has joined #zuul | 20:41 | |
| *** sdake has quit IRC | 20:44 | |
| *** sdake has joined #zuul | 20:47 | |
| *** sdake has quit IRC | 21:01 | |
| *** sdake has joined #zuul | 21:02 | |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Add provides/requires support https://review.openstack.org/633605 | 21:04 |
| *** sdake has quit IRC | 21:38 | |
| *** sdake has joined #zuul | 21:49 | |
| *** openstackgerrit has quit IRC | 21:50 | |
| *** ianw_pto is now known as ianw | 21:59 | |
| *** openstackgerrit has joined #zuul | 22:07 | |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul master: Add provides/requires support https://review.openstack.org/633605 | 22:07 |
| *** themroc has quit IRC | 22:31 | |
| corvus | tobiash, SpamapS: ^ that is green. i know it *looks* like a big change, but there's actually only 189 lines of new code (46 of which is a sql migration). the interesting bits are all in model.py. the remaining 830 lines are testing, docs, and re-organization. mostly testing. | 23:00 |
| *** sdake has quit IRC | 23:01 | |
| *** rlandy is now known as rlandy|bbl | 23:24 | |
| *** sdake has joined #zuul | 23:29 | |
| *** sdake has quit IRC | 23:33 | |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul-jobs master: Allow stage-output to copy root-owned files https://review.openstack.org/634069 | 23:39 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!