| *** alumno has joined #zuul | 04:08 | |
| *** alumno has quit IRC | 04:32 | |
| *** hhcmralgsp has joined #zuul | 06:37 | |
| *** hhcmralgsp has quit IRC | 06:38 | |
| *** xoqbebvbqj has joined #zuul | 07:16 | |
| *** xoqbebvbqj has quit IRC | 07:16 | |
| *** hashar has joined #zuul | 07:49 | |
| *** smyers has quit IRC | 09:02 | |
| *** smyers has joined #zuul | 09:04 | |
| *** xinliang has quit IRC | 10:30 | |
| *** xinliang has joined #zuul | 10:42 | |
| *** xinliang has joined #zuul | 10:42 | |
| *** dmsimard is now known as dmsimard|afk | 11:05 | |
| *** bhavik1 has joined #zuul | 12:00 | |
| *** bhavik1 has quit IRC | 12:13 | |
| *** hashar has quit IRC | 12:37 | |
| *** hashar has joined #zuul | 13:59 | |
| openstackgerrit | Merged openstack-infra/zuul-jobs master: add more debugging to the upload-pypi role https://review.openstack.org/514394 | 14:48 |
|---|---|---|
| *** dmsimard|afk is now known as dmsimard | 15:30 | |
| openstackgerrit | Merged openstack-infra/zuul feature/zuulv3: Zuul: add file extension to playbook path https://review.openstack.org/516135 | 16:10 |
| openstackgerrit | Merged openstack-infra/zuul feature/zuulv3: Use cached branches on tenant reconfiguration https://review.openstack.org/515971 | 16:10 |
| *** bhavik1 has joined #zuul | 16:33 | |
| *** bhavik1 has quit IRC | 16:41 | |
| * SpamapS reads the section of docs about shared queues | 16:50 | |
| Shrews | Do I really need a company-mandated training course to know that "different people have different ideas". Geez | 17:10 |
| * Shrews grumbles under his breath | 17:10 | |
| *** hashar is now known as hasharAway | 17:16 | |
| *** harlowja has joined #zuul | 17:19 | |
| *** hasharAway has quit IRC | 17:23 | |
| dmsimard | Shrews: the ethics thing? heh | 17:24 |
| dmsimard | Shrews: you'd be surprised about what is common sense for you and me isn't for everyone and it's stuff that's more or less required being in a publicly traded company and all | 17:25 |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul feature/zuulv3: Fix syntax with gear unRegisterFunction() https://review.openstack.org/516403 | 17:26 |
| pabelanger | jeblair: SpamapS: ^should fix our governor thread. | 17:27 |
| pabelanger | Not sure the best way to test that in gate, maybe some way we can mock load | 17:27 |
| *** hashar has joined #zuul | 17:28 | |
| jeblair | pabelanger: yeah, this is a case where i would be happy to have a test which mocked out os.getloadavg() and did a 'status' call to gearman to verify that the function was unregistered | 17:29 |
| pabelanger | yah | 17:30 |
| SpamapS | how did that slip through? | 17:30 |
| SpamapS | has the executor just been crashing at high load? | 17:30 |
| pabelanger | not sure about crashing | 17:30 |
| SpamapS | seemed like it was working so.. ? | 17:30 |
| pabelanger | I just seen the loadavg on http://grafana.openstack.org/dashboard/db/zuul-status | 17:30 |
| pabelanger | then looked | 17:31 |
| jeblair | just there's an exception handler, so it's just no-opping | 17:31 |
| pabelanger | but ianw did restart zuul this morning and noticed the exception | 17:31 |
| jeblair | pabelanger: did you want to write the test, or should i take a stab at it? | 17:32 |
| jeblair | SpamapS: i'm guessing now that we have 10 executors, we're just dealing with the load naturally most of the time. | 17:33 |
| pabelanger | jeblair: Yah, I don't mind give it a try. But if going to hold up 516403 maybe you want to take the stab | 17:33 |
| SpamapS | jeblair: but at the time we wrote it.. it was unregistering things in the log.. so weird. | 17:34 |
| jeblair | pabelanger: i think we can merged 403 without it. it's all yours. let me know if i can help. | 17:34 |
| pabelanger | k | 17:34 |
| SpamapS | I can write the test too | 17:36 |
| SpamapS | since it's my code originally | 17:36 |
| jeblair | SpamapS: the original version did not have an exception handler, so it would have emitted a log line saying "unregistering", then hit the error and the thread would have died | 17:36 |
| SpamapS | awesome | 17:37 |
| SpamapS | pabelanger: were you writing a test then? | 18:37 |
| pabelanger | SpamapS: I am, but haven't started just yet. Working on logstash-workers atm | 18:38 |
| pabelanger | SpamapS: https://review.openstack.org/516403/ isn't +A so, if you want to add to it or append as follow up, wfm | 18:39 |
| pabelanger | or we can just land and I was going to work on it this evening | 18:39 |
| SpamapS | pabelanger: I'm going to poke at it now | 18:40 |
| SpamapS | I actually recall wanting to write one and then just pushing forward because everything was on fire | 18:40 |
| openstackgerrit | Andreas Jaeger proposed openstack-infra/zuul feature/zuulv3: Remove zuul-migrate job https://review.openstack.org/516028 | 18:40 |
| jeblair | SpamapS: 18:47 < openstackgerrit> James E. Blair proposed openstack-infra/gear master: server: make stats more efficient https://review.openstack.org/516422 | 18:47 |
| * SpamapS looks | 18:48 | |
| jeblair | SpamapS: our logstash gearman queue is 600k entries long, which made me realize geard can totally handle that, because everything related to queue lengths is O(1), except those stats calculations. | 18:48 |
| jeblair | (which are O(n)) | 18:49 |
| SpamapS | jeblair: I"m surprised to year that the workers stat isn't all that useful. | 18:50 |
| SpamapS | s/year/hear/ | 18:50 |
| SpamapS | but I suppose if it's a stable worker set that makes sense. | 18:50 |
| jeblair | SpamapS: i think the reason i say that is because it's only an aggregate count of workers. that's useful in some cases, but others you really need to know workers by function. | 18:51 |
| jeblair | (basically, it would be useful if you have a dynamic set of workers all with the same functions) | 18:52 |
| jeblair | SpamapS: i'm happy to keep it if you think it's valuable. i don't have an idea of how to calculate it without looping though (though generally that should be a pretty fast loop) | 18:54 |
| jeblair | SpamapS: okay, i think i have an idea :) | 18:58 |
| SpamapS | jeblair: yeah workers by function is far far far more useful | 19:01 |
| SpamapS | total workers is a bit of a corner case | 19:01 |
| SpamapS | Oh interesting | 19:02 |
| SpamapS | jeblair: gear has an OS X bug | 19:02 |
| SpamapS | File "/Users/cbyrum/src/openstack-infra/zuul/.tox/py36/lib/python3.6/site-packages/gear/__init__.py", line 2698, in Server | 19:02 |
| jeblair | meh, my idea is hard. i'm going to leave the change as-is for now (dropping workers) | 19:02 |
| SpamapS | edge_bitmask = select.EPOLLET | 19:02 |
| SpamapS | AttributeError: module 'select' has no attribute 'EPOLLET' | 19:03 |
| * SpamapS curses this stupid mac | 19:03 | |
| jeblair | SpamapS: oh, yeah, geard doesn't work on osx. gear client does though. | 19:03 |
| SpamapS | can't run unit tests for zuul because of it tho | 19:03 |
| * SpamapS deals | 19:03 | |
| Shrews | SpamapS: yeah, that particular bit of mac-osx-iness was the final straw (of many) that drove me away from ever using a mac for dev again | 19:06 |
| Shrews | i know there's a workaround with kqueue, but it just kept popping up | 19:08 |
| SpamapS | Shrews: I suppose if we make it work on OS X we get *BSD for free. | 19:22 |
| SpamapS | but I find myself working in my Ubuntu VM more and more | 19:22 |
| SpamapS | if nothing else, the vim settings are better | 19:22 |
| SpamapS | hrm proving harder than I thought.. have to monkey patch stuff before the executor starts | 20:07 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Add support for override-checkout, deprecate override-branch https://review.openstack.org/516451 | 21:11 |
| SpamapS | you know, I used to think zuul was about scaling to a massive team... | 21:50 |
| SpamapS | It's pretty awesome just using it for 2 people. | 21:50 |
| SpamapS | jeblair: zuul meeting today.. I'm double booked for the first 30min. But would like to discuss turning our sights on 3.0 soon... possibly targetting a release around PTG-ish time. | 21:53 |
| jeblair | SpamapS: ack. i plan on sending out an email with the priorities list we brainstormed at the last ptg today or tomorrow. i think that'll be good to help us see the path to get there. i agree that timeline (feb) is plausible. | 21:57 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul feature/zuulv3: Add support for override-checkout, deprecate override-branch https://review.openstack.org/516451 | 22:00 |
| jeblair | it's zuul meeting time in #openstack-meeting-alt | 22:00 |
| SpamapS | jeblair: awesome. I hope to have enough capital and data from my early experiments to make a strong push for some GoDaddy teams to adopt it fully (otherwise we'll probably standardize on Jenkins Pipeline and groovy.. blech) | 22:00 |
| *** hashar has quit IRC | 22:11 | |
| openstackgerrit | Ian Wienand proposed openstack-infra/zuul feature/zuulv3: Check start time for wait_time key https://review.openstack.org/516465 | 22:29 |
| jeblair | the override-checkout change is ready for review: https://review.openstack.org/516451 | 23:03 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!