| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2024-02-08 | |||
| 18:18:05 | chris218 | So devstack is the way? | |
| 18:18:30 | sean-k-mooney | nova has extenstive unit and functional tests | |
| 18:18:43 | sean-k-mooney | but if you want more then that you shoudl use devstack or some other tool | |
| 18:18:48 | chris218 | I was recently playing with the idea of writing my own driver and was wondering what would be the best way to test it | |
| 18:19:31 | sean-k-mooney | we offically do not supprot out of tree drivers and its unlikely we will add a new intree dirver in the medium to long term | |
| 18:19:48 | sean-k-mooney | what hypervior were you hoping to integrate with | |
| 18:20:36 | sean-k-mooney | there is not sable external api for writing a virt driver for nova | |
| 18:20:49 | chris218 | It's mostly me trying to practice working with python in a larger codebase, but either cloud hypervisor or qemu through qmp instead of libvirt | |
| 18:21:14 | sean-k-mooney | :) i have breifly looked at cloud hypervisor | |
| 18:21:31 | sean-k-mooney | if you wanted it to produced functional vms then devstack is the best approch | |
| 18:21:45 | chris218 | Awesome thanks for your help | |
| 18:22:44 | sean-k-mooney | on comment i would make is if you wanted ot gain experiance with workign on a large project and its not for work | |
| 18:23:09 | sean-k-mooney | there is alwasy a large bug backlog if you felt so inclidned | |
| 18:23:54 | sean-k-mooney | chris218: but if you wanted a refecne fo what it might look like this is what an out of tree driver would look like https://github.com/cdent/etcd-compute | |
| 18:24:22 | sean-k-mooney | actuly thats not a good example | |
| 18:24:40 | sean-k-mooney | that and example of a driver that woudl remove the rpc bus and use etcd instead | |
| 18:25:24 | chris218 | Oh awesome thanks, I was wondering why Kolla has etcd option | |
| 18:25:48 | sean-k-mooney | its not related to this | |
| 18:25:54 | chris218 | Oh :( | |
| 18:25:58 | sean-k-mooney | that was just a tought experiment that never went anywhere | |
| 18:26:19 | sean-k-mooney | ectd in kolla is for use as a cache backend or distributed lock manager | |
| 18:26:42 | sean-k-mooney | cinder via tooz can use etcd and i belive that is what koolla supprot it for | |
| 18:26:57 | sean-k-mooney | oslo.cache i belive also supprots etcd | |
| 18:27:14 | sean-k-mooney | chris218: if your familar with kolla | |
| 18:27:20 | sean-k-mooney | kolla has a dev mode | |
| 18:27:34 | sean-k-mooney | where in bindmoutns local git repos into the conatiner | |
| 18:27:39 | chris218 | Btw while we're talking I was wondering for a long time, why do openstack projects still use pbr and setup.cfg? Don't setuptools with pyproject.toml have everything that pbr wanted to add? | |
| 18:27:48 | chris218 | Ooh that sounds nice | |
| 18:28:00 | sean-k-mooney | short version no | |
| 18:28:06 | sean-k-mooney | there are several feature gaps | |
| 18:28:40 | sean-k-mooney | pytproject.toml also changes how you manage depencies and we dont have a good solution for upper-constraits.txt | |
| 18:28:48 | sean-k-mooney | https://github.com/openstack/requirements/blob/master/upper-constraints.txt | |
| 18:29:03 | sean-k-mooney | so we dont have a good solution for how to maintain co installablity between all the repos | |
| 18:29:27 | sean-k-mooney | with that said it does not mean we cant add pyproject.toml | |
| 18:29:33 | chris218 | Oh I didn't even know that existed | |
| 18:29:33 | sean-k-mooney | i actully started doing that in some places | |
| 18:29:47 | sean-k-mooney | but we cant just drop pbr right now | |
| 18:30:22 | sean-k-mooney | chris218: https://review.opendev.org/q/topic:%22pip-23.1-support%22 | |
| 18:30:52 | sean-k-mooney | chris218: one of the things we depend on pbr for | |
| 18:32:53 | chris218 | But I think it should be possible to drop it from runtime deps and use importlib to check version, no? | |
| 18:35:33 | sean-k-mooney | is we make sue of a custome setuptools entry point that generage wsgi scripts for the rest apis | |
| 18:35:33 | sean-k-mooney | wehn we enabel pyproject.toml even with pbr that breaks if you do an editiable install | |
| 18:35:33 | sean-k-mooney | so to progress with removing pbr we need to complete https://review.opendev.org/q/topic:%22remove-wsgi_scripts%22 | |
| 18:35:33 | sean-k-mooney | bauzas: speaking of https://review.opendev.org/q/topic:%22remove-wsgi_scripts%22+repo:openstack/nova can you review that | |
| 18:35:33 | sean-k-mooney | it would be good to land the first 2 patchs | |
| 18:37:03 | sean-k-mooney | chris218: am we dont really have a runtim dep on setuptools | |
| 18:37:17 | sean-k-mooney | at least not in most of the code | |
| 18:37:22 | sean-k-mooney | but we do use stevador | |
| 18:37:41 | sean-k-mooney | https://github.com/openstack/stevedore | |
| 18:37:50 | sean-k-mooney | or rather https://opendev.org/openstack/stevedore | |
| 18:41:23 | chris218 | Hmm so openstack has everything home grown 😅 | |
| 18:42:12 | sean-k-mooney | not entirly openstack predates a lot of the modern python tooling | |
| 18:42:25 | sean-k-mooney | pbr is just a wrapper around setup tools | |
| 18:43:08 | chris218 | Yea like having own api-ref system instead of OpenAPI since it's the "new" hot thing | |
| 18:43:35 | sean-k-mooney | its not about it being the new hot thing | |
| 18:43:48 | sean-k-mooney | we actully discused supproting openAPI a few years ago | |
| 18:44:03 | sean-k-mooney | but we did nto have contibutors that coudl work on it | |
| 18:44:25 | artom | Nobody here is hot or new :D | |
| 18:44:33 | chris218 | Haha | |
| 18:44:41 | chris218 | I'm new hello | |
| 18:45:05 | sean-k-mooney | the api-ref is part of our docs by the way where as openapi is a specification for defining api and automatic client generation form those schemas | |
| 18:46:02 | sean-k-mooney | we have json schemas for exmaple in our apis already but we never had the time to rewirte themn to and openapi compaitble format | |
| 18:46:25 | chris218 | Well you can also generate docs from it, I've seen one project of person from sdk team working on it privately | |
| 18:47:04 | sean-k-mooney | right i knwo there has been some discussion about it in the context of the opsntack sdk | |
| 18:47:19 | sean-k-mooney | we really dont wnat the sdk to create there own scemas | |
| 18:47:26 | sean-k-mooney | that woudl be a very bad thing | |
| 18:48:13 | sean-k-mooney | stephenfin has been trying to push the converstaion back to adding openapi supprot to the phyton projects instead | |
| 18:48:33 | sean-k-mooney | i.e. adding openapi supprot to nova/neutron ectra | |
| 18:49:14 | chris218 | Ok last question and I'll stop nagging you, what's OpenStack stance on using from __future__ import annotations? | |
| 18:49:39 | sean-k-mooney | we have to main python compatibality with 3.8 currrently | |
| 18:49:46 | sean-k-mooney | that will raise to 3.9 soon | |
| 18:50:07 | sean-k-mooney | i would gnerally say from __future__ import <anything> shoudl not be used | |
| 18:50:24 | sean-k-mooney | unless thte feature has been fully merged in a released python | |
| 18:51:19 | sean-k-mooney | we defien our testing runtimes here https://github.com/openstack/governance/blob/master/reference/runtimes/2024.1.rst | |
| 18:51:27 | sean-k-mooney | one file per release | |
| 18:53:02 | sean-k-mooney | chris218: im not exactly show what from __future__ import annotations is provideing | |
| 18:53:33 | sean-k-mooney | we do use typing and mypy to a limited degree | |
| 18:53:38 | chris218 | It turns all type hints into strings so you don't need to import their dependencies | |
| 18:54:03 | sean-k-mooney | right but you dont need that anymore | |
| 18:54:18 | sean-k-mooney | we have a mix of string and non string import | |
| 18:54:24 | sean-k-mooney | *type hints | |
| 18:54:35 | chris218 | True I guess 🤔 | |
| 18:54:50 | sean-k-mooney | this was part of https://peps.python.org/pep-0563/ right | |
| 18:55:13 | sean-k-mooney | in general i prefer not to use the string version but somethiems i will to avoid the import | |
| 18:55:24 | sean-k-mooney | the string version is more error prone | |
| 18:55:56 | sean-k-mooney | https://codesearch.opendev.org/?q=from+__future__+import+annotations&i=nope&literal=nope&files=&excludeFiles=&repos= | |
| 18:56:24 | sean-k-mooney | it is used but i think it was mostly for the tanstion period when we had to supprot python 2.7 | |
| 18:57:08 | chris218 | Good to know thanks for the clarification. Also I didn't know codesearch existed that looks awesome | |
| 18:57:31 | sean-k-mooney | its an instance of hound | |
| 18:57:45 | sean-k-mooney | https://github.com/hound-search/hound | |
| 18:59:54 | opendevreview | melanie witt proposed openstack/nova master: WIP Support encrypted backing files for qcow2 https://review.opendev.org/c/openstack/nova/+/907961 | |
| 18:59:54 | opendevreview | melanie witt proposed openstack/nova master: libvirt: Introduce support for raw with LUKS https://review.opendev.org/c/openstack/nova/+/884313 | |
| 18:59:55 | opendevreview | melanie witt proposed openstack/nova master: libvirt: Introduce support for rbd with LUKS https://review.opendev.org/c/openstack/nova/+/889912 | |
| 19:03:48 | opendevreview | VO LE HUY proposed openstack/nova stable/2023.2: Force death and notice that the system clock is unstable https://review.opendev.org/c/openstack/nova/+/908519 | |
| 19:06:26 | opendevreview | VO LE HUY proposed openstack/nova stable/2023.2: Force down and logging that the system clock is unstable so administrator need to time sync by ntpd or chronyc, ... https://review.opendev.org/c/openstack/nova/+/908519 | |
| 19:20:53 | opendevreview | VO LE HUY proposed openstack/nova stable/2023.2: Do not use the abs(elapsed) function when elapsed is a negative number. Wrong operating principle but nova service still shows up status. https://review.opendev.org/c/openstack/nova/+/908519 | |
| #openstack-nova - 2024-02-09 | |||
| 00:35:29 | opendevreview | VO LE HUY proposed openstack/nova master: Shouldn't use the abs(elapsed) function when elapsed is a negative number. Wrong operating principle but nova service still shows up status. https://review.opendev.org/c/openstack/nova/+/908536 | |
| 02:03:28 | opendevreview | Takashi Kajinami proposed openstack/nova master: libvirt: Detect vtpm support by libvirt https://review.opendev.org/c/openstack/nova/+/908508 | |
| 02:46:04 | opendevreview | Takashi Kajinami proposed openstack/nova master: libvirt: Detect vtpm support by libvirt https://review.opendev.org/c/openstack/nova/+/908508 | |
| 03:01:40 | opendevreview | Takashi Kajinami proposed openstack/nova master: libvirt: Ensure both swtpm and swtpm_setup exist for vTPM support https://review.opendev.org/c/openstack/nova/+/908541 | |