| zbr | dmsimard: how reliable is mitogen these days? | 11:58 |
|---|---|---|
| zbr | i was wondering If I could try to use it during molecule functional tests,... without breaking them. | 11:59 |
| zbr | ouch: Ansible 2.8 interpreter discovery and become plugins are not yet supported. | 12:00 |
| dmsimard | zbr: I do not currently use it on a regular basis so can't quite vouch if it's reliable or not | 13:53 |
| dmsimard | But I get questions about if it works for ara so I figured I'd write a small job for it | 13:54 |
| dmsimard | Same for awx, I'll have a test job for it soon | 13:54 |
| dmsimard | Oh, and ansible-runner too. | 13:54 |
| sshnaidm | zbr, should be supported for 2.8 | 14:12 |
| zbr | sshnaidm: unusable due to https://github.com/dw/mitogen/issues/597 -- vote/comment. | 14:22 |
| zbr | it faile don the first playbook i tried, w/o python detection is of no use to me | 14:22 |
| sshnaidm | zbr, hmm.. that was told they fixed this a half year before | 14:23 |
| sshnaidm | maybe a new issue | 14:23 |
| zbr | well, not really. and not even related to rhel-8, it failed on my bsd router, with has standard python | 14:23 |
| sshnaidm | zbr, did you try with 2.8 or 2.9? | 14:23 |
| sshnaidm | bsd? not really standard.. :) | 14:24 |
| zbr | for routers, ojhh. yeah. | 14:24 |
| zbr | i think that >90% of networking is running on BSD-variants | 14:25 |
| zbr | i wanted to switch to linux distro, but there is none that would have even a mediocre web interface | 14:26 |
| sshnaidm | zbr, cockpit? | 14:27 |
| sshnaidm | https://cockpit-project.org/ | 14:27 |
| zbr | what? not doing official work today | 14:27 |
| sshnaidm | other cockpit :D | 14:28 |
| zbr | uhh, bit releaved. | 14:28 |
| zbr | that is not a router-os, more of "web version of glances" | 14:29 |
| sshnaidm | I think any distro can use it' | 14:30 |
| zbr | sshnaidm: i was interested to try mitogen with molecule functional tests, to speed them up. you are welcome to take the challenge. | 14:30 |
| sshnaidm | zbr, if ansible is not the bottleneck there, it won't help much | 14:31 |
| sshnaidm | zbr, I tried mitogen with oooq and tripleo | 14:31 |
| zbr | it is a signifiant part of it, maybe even >50%. | 14:31 |
| sshnaidm | zbr, ok, then maybe will work, it's great when a lot of small and short ansible tasks | 14:32 |
| sshnaidm | zbr, it's actually 2 env vars to set | 14:33 |
| zbr | yep. and then you fail to detect python | 14:34 |
| sshnaidm | zbr, and if explicit ansible_python_interpreter? | 14:34 |
| zbr | and I have no plans to hardcode path for each image | 14:34 |
| zbr | probably it will work, but didn't bother | 14:34 |
| sshnaidm | worth to check anyway, if it helps at all | 14:34 |
| sshnaidm | and worth an effort | 14:35 |
| zbr | maybe, i preferred to implement other fixes. | 14:36 |
| zbr | focusing on longest func tests, one by one. | 14:36 |
| zbr | sadly zuul performance does vary between jobs making harder to measure outcomes. locally i have more reliable results. | 14:37 |
| zbr | i bet with few days of work, I could bring functional duration to ~50% of what it was. | 14:37 |
| zbr | now it takes at least 3h to gate a change, 1.5 check, 1.5 gate and I do not want to do sharding. | 14:38 |
| sshnaidm | what does take 1.5h? | 14:39 |
| zbr | https://github.com/ansible-community/molecule/pull/2510 -- mainly functionals are running after the quick ones | 14:41 |
| zbr | quick ones are <15min, functionals are >1h, usually ~1:15 | 14:41 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!