*** pcaruana has joined #zuul | 10:59 | |
*** pcaruana has quit IRC | 11:00 | |
*** bhavikdbavishi has joined #zuul | 11:25 | |
*** bhavikdbavishi has quit IRC | 12:02 | |
mordred | SpamapS: we really need to finish refactoring that streaming system | 14:55 |
---|---|---|
SpamapS | mordred: what's next on that todo? | 14:58 |
SpamapS | mordred: the refactor I did certainly made it easier to read.. but is there more? | 14:58 |
mordred | SpamapS: this is the server-side bit ... | 14:58 |
SpamapS | ah yeah | 14:58 |
SpamapS | so you can signal "that's really the end of the build" vs. "bbbbuuuhh bye" | 14:59 |
mordred | https://review.openstack.org/#/q/topic:zuul-stream-rework | 14:59 |
mordred | yeah | 14:59 |
SpamapS | neat | 15:00 |
mordred | I also need to go dig in to the stuff bcoca has been doing in ansible upstream | 15:00 |
SpamapS | would be really cool to have streaming built in to ansible. :) | 15:10 |
SpamapS | mordred: how sweet would it be if the console worked like a ring buffer, and it always just got saved into the builds table? | 15:14 |
SpamapS | would be a convenient answer to "ugh, and I have to set up log serving?" | 15:15 |
mordred | heh | 15:15 |
mordred | well - I mean - unfortunately, console logs being stored into the builds table would kill us in openstack land | 15:16 |
* mordred waves to devstack | 15:16 | |
SpamapS | Even 25 lines you think? | 15:16 |
mordred | I may not be following the thing you're saying :) | 15:17 |
SpamapS | gzip it up, you can probably fit it into 255 bytes. | 15:17 |
mordred | oh golly no - these bad boys are huge | 15:17 |
SpamapS | ring buffer, meaning, only show the last X lines. | 15:17 |
SpamapS | It's not "the log", it's "what's happening" | 15:17 |
mordred | yeah - I think that would still kill us - because now we'd have to have a scalable mysql service taking a gajillion writes and deletes :( | 15:18 |
SpamapS | oh I wouldn't put the ring buffer in mysql | 15:18 |
SpamapS | just at the end | 15:18 |
mordred | at the end it would just save the last 25 lines into mysql? | 15:18 |
SpamapS | Just have the executor hit the finger service at the end. | 15:18 |
SpamapS | The place I'd do the ring buffer is in the viewer. | 15:19 |
SpamapS | to keep it lightweight | 15:19 |
SpamapS | but then if you request a build that isn't running now, it could still show you the last 25 lines | 15:19 |
SpamapS | maybe with the log link at the bottom | 15:19 |
mordred | I think I need more coffee - cause I'm just not following you but I really want to | 15:20 |
SpamapS | It's mostly that I don't like how different viewing "logs" is from viewing the streams. | 15:20 |
mordred | oh - well, *that* is just a point in time bug | 15:20 |
SpamapS | and I def need more caffeine as this isn't really all that awesome. :-P | 15:20 |
mordred | the working idea there is that once the build page exists in the dashboard, then it can load the console_log.json (or console_log.yaml) file from the link provided to it by the zuul api - and render it appropriately - so that for the console log you're *always* going through the dashboard, and if it's live, it'll be live, and if it's finished it'll be the log data | 15:22 |
mordred | and then, like you said, links to more log files, etc | 15:23 |
mordred | SpamapS: but you know - what should probably happen is that we should write something down describing the future, you know, like a spec or something | 15:24 |
mordred | cause I think right now it mostly exists as memories of conversations | 15:24 |
SpamapS | hrm. wtf http://paste.openstack.org/show/739415/ | 15:24 |
SpamapS | mordred: memories are only specs for about 2 months. ;) | 15:25 |
mordred | SpamapS: ugh. | 15:25 |
mordred | SpamapS: so - there are a couple of possibilities there | 15:25 |
SpamapS | I have the same deb.. it installs /usr/bin/npm ... :-P | 15:25 |
mordred | is it possible this decided to install from distro rather than upstream? | 15:26 |
* mordred reading code | 15:26 | |
SpamapS | could be | 15:26 |
SpamapS | 18.04's might look newer | 15:26 |
SpamapS | oh | 15:26 |
mordred | yeah | 15:26 |
SpamapS | node_version: 6 | 15:27 |
* SpamapS will poke 8 into that | 15:27 | |
mordred | I bet it does - we should add an apt-pin to pull from the nodesource versions -- you can pin to a source right? | 15:27 |
mordred | 2018-12-30 15:15:18.253193 | builder | v8.10.0 | 15:27 |
mordred | SpamapS: ^^ you installed version 8 though | 15:27 |
mordred | oh - but it would have added the v6 repo - got it | 15:27 |
mordred | we should still add an apt pin | 15:27 |
mordred | SpamapS: it's pin: origin deb.nodesource.com I think yeah? | 15:28 |
* SpamapS fetches the pinning scrolls | 15:28 | |
SpamapS | mordred: correct | 15:29 |
mordred | k working on patch | 15:30 |
openstackgerrit | Monty Taylor proposed openstack-infra/zuul-jobs master: Pin nodejs package to upstream repo https://review.openstack.org/627823 | 15:34 |
mordred | SpamapS: ^^ | 15:34 |
openstackgerrit | Monty Taylor proposed openstack-infra/zuul master: Add nodesets API route https://review.openstack.org/605877 | 15:39 |
*** sshnaidm_ is now known as sshnaidm | 15:44 | |
mordred | SpamapS: if you get bored, there's a big pile of things with 1 +2 at https://review.openstack.org/#/q/topic:react-zuul that would benefit from a second set of eyes - especially since you were just hacking in that area | 15:49 |
SpamapS | Oh yes, weirdly, I've reached intermediate level react the last few months. | 15:49 |
* SpamapS waits for gertty to get done thinking | 15:49 | |
SpamapS | 2018-12-30 15:30:29,257 INFO zuul.ExecutorServer: Unregistering due to high system load 5.13427734375 > 5.0 | 16:04 |
SpamapS | 2018-12-30 15:33:09,264 INFO zuul.ExecutorServer: Re-registering as job is within its limits 4.77587890625 <= 5.0, 26.1% <= 5.0, running, 78.7% <= 5.0, 0 <= 4.0 | 16:04 |
SpamapS | yaaaay | 16:04 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!