| === benj_8 is now known as benj_ | ||
| blackboxsw | falcojr: I'm finally going through your pr, will have minor unrelated suggestions on https://github.com/canonical/pycloudlib/pull/148/files. Just testing to confirm | 16:07 |
|---|---|---|
| blackboxsw | falcojr: also I know I'm way late to this discussion and we've brought it up in standup multiple times, this branch is essentially doing away with a restart primitive, and making "restart" just a wrapper for shutdown(wait=True)/start. Generally, I don't think this really matters, but it does exercise a slightly different path on a platform. shutdown/start on a platform != restart. So, we could be exposed to something there on a | 16:17 |
| blackboxsw | platform by not using the discrete restart operation on a given cloud. | 16:17 |
| falcojr | blackboxsw: right, I'm out backing that part out (I intentionally did it as a separate commit) if we think that's going to cause problems | 16:18 |
| blackboxsw | on first blush, I don't really honestly believe its a problem, but it's a blind spot in pycloudlib cloud operations if we don't use the cloud/platform-specific restart operation. | 16:18 |
| blackboxsw | falcojr: " I'm out backing that part out" is not parsing for me | 16:19 |
| blackboxsw | *ok*? or *not* | 16:19 |
| blackboxsw | :) | 16:19 |
| blackboxsw | maybe paride also has more context on the ec2instance.restart in pycloudlib as he most recently extended the error handling there. | 16:20 |
| blackboxsw | I have a suggestion for lxc.restart if were keeping platform-specific restart logic intact where applicable | 16:21 |
| falcojr | blackboxsw: sorry, meant to say "I'm fine backing that part out" | 16:26 |
| shalok | Is there a comprehensive reference for the cloud-config? I've found lots of examples, but these don't appear to use all of the supported configs. | 17:05 |
| falcojr | shalok: https://cloudinit.readthedocs.io/en/latest/topics/modules.html should be what you're looking for | 17:06 |
| blackboxsw | shalok: other significant examples are @ https://cloudinit.readthedocs.io/en/latest/topics/examples.html | 18:05 |
| shalok | Yeah, I've mostly been relying on those examples. The modules docs is more what I was looking for. | 18:06 |
| blackboxsw | falcojr: trying to work on the patch for your pycloudlib branch. about ~30 mins left. I'm re-running ua-client integration tests on lxc to see performance improvement/check for failures | 18:40 |
| blackboxsw | falcojr: done review posted https://github.com/canonical/pycloudlib/pull/148#pullrequestreview-676584230 | 19:05 |
| blackboxsw | see what you think | 19:05 |
| powersj | https://christine.website/blog/cloud-init-2021-06-04 | 20:51 |
| Xe | powersj: that's my article yeah! | 20:57 |
| powersj | awesome - thanks for the write up. :) I like coming across folks who are trying out cloud-init and see where we can do better | 20:58 |
| Xe | the documentation could use some love, but i'm probably doing an obscure thing | 20:58 |
| powersj | I think a good majority of folks are interacting with cloud-init via a cloud itself and less so via qemu, but that said I wrote a post on this very thing last week after co-workers asked me to: https://powersj.io/posts/ubuntu-qemu-cli/ so I woulnd't say it is too obscure :) | 21:01 |
| Xe | oh, that post | 21:03 |
| Xe | i used it as reference when i was getting stuff working for the first time! | 21:03 |
| powersj | ha :) | 21:05 |
| haitch | "cloud-init analyze blame" gives me multiple boot record for a single boot | 22:24 |
| haitch | is this common and anyone know how can I troubleshoot this? | 22:25 |
| rharper | haitch: not common, but in the past we've had some log formatting issues on different OSes which have cloud-init installed but different logging configurations for cloud-init configured. If you can, use paste.ubuntu.com to capture your output, and maybe /var/lib/cloud-init.log ; boot records are found by looking for the various times with cloud-init is invoked; so if the system is running cloud-init init command outside of | 22:32 |
| rharper | booting, you may see more records as they get appended to the log file that's parsed by cloud-init analyze | 22:32 |
| haitch | https://paste.ubuntu.com/p/SZ6xsRzKC4/ | 22:56 |
| haitch | I confirmed just have 1 real VM boot | 22:57 |
| rharper | oh, I know what this is .. | 23:20 |
| rharper | on azure, they pre-provision VMs, so they're launched and then paused, then when they're needed, they wake up and resume the rest of boot; that said, I thought we had a bug to fix the parsing of cloud-init.log for analyze ... it'd be good to file one just to track the issue; | 23:21 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!