| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2021-06-09 | |||
| 13:13:51 | sean-k-mooney | which is what the ssh lib we use | |
| 13:13:56 | stephenfin | correct | |
| 13:14:13 | sean-k-mooney | hum its not really a test-requireemnt is it | |
| 13:14:27 | sean-k-mooney | oh for the mypy job? | |
| 13:14:27 | stephenfin | it is. mypy isn't a runtime dependency | |
| 13:14:33 | sean-k-mooney | ok | |
| 13:15:19 | sean-k-mooney | stephenfin: did you ever add mypy to pre-commit? | |
| 13:15:37 | stephenfin | not yet, no. Just haven't set aside the time to figure it out | |
| 13:15:46 | sean-k-mooney | no worries | |
| 13:16:19 | gibi | I cannot reproduce it with pep8 env either. but it seem my machine installs a bit older mypy than the gate job | |
| 13:19:10 | tosky | melwitt: I think there is a chicken-and-egg problem with the backports, you may either need to squash them (but maybe that would be too much) or set the failing legacy job in the first patch to non-vpting, to have it replaced with a native one with one of the newer patches | |
| 13:19:30 | gibi | stephenfin: OK it needs a new mypy version to | |
| 13:19:31 | gibi | o | |
| 13:19:40 | lyarwood | amodi: updated, can you build a devstack env and get this passing before you push another revision? | |
| 13:20:09 | gibi | stephenfin: it does not fail with mypy==0.812 but it fails with mypy==0.901 | |
| 13:20:51 | stephenfin | hmm, good spot. I saw the release date on types-paramiko was pretty recent so I figured that was it | |
| 13:20:59 | stephenfin | I'll see if I can figure out what changed | |
| 13:21:19 | lyarwood | so we could cap mypy to unblock the gate instead right? | |
| 13:21:20 | amodi | lyarwood: um, i havent deployed one in a long time and how would i deploy with these parameters, ill hv to figure out | |
| 13:21:43 | sean-k-mooney | lyarwood: ya that would work if we use UC for it | |
| 13:22:31 | stephenfin | lyarwood: we'd have to cap it | |
| 13:22:33 | stephenfin | no uncap | |
| 13:22:37 | stephenfin | the higher version is the issue | |
| 13:22:49 | sean-k-mooney | its in test-requrimets not installed in the tox.ini | |
| 13:23:07 | sean-k-mooney | so if we cap it in upper-constratints in the requirements repo it shoudl work | |
| 13:23:18 | gibi | stephenfin: "The upcoming switch to modular typeshed in mypy 0.900" | |
| 13:23:24 | gibi | that was hit us | |
| 13:23:27 | stephenfin | ahhh | |
| 13:23:32 | kashyap | stephenfin: Saw your review on that display change. I'll respond in a bit | |
| 13:23:36 | kashyap | Thank you | |
| 13:23:38 | stephenfin | wonderful \o/ | |
| 13:23:46 | gibi | https://mypy-lang.blogspot.com/2021/05/the-upcoming-switch-to-modular-typeshed.html | |
| 13:24:03 | gibi | we are lucky that only one thing was missing so far :) | |
| 13:24:32 | sean-k-mooney | :( non nova type definition we can steal | |
| 13:26:10 | sean-k-mooney | i wonder is there a way we can tell it to ignore third party libs if the hints are not present | |
| 13:26:49 | sean-k-mooney | i.e. can we supress the error and maybe make ti a warning | |
| 13:29:19 | sean-k-mooney | "You can also now run mypy --install-types to install all missing stub packages in the mypy run (or the previous run, if you don’t pass any files to check): " | |
| 13:29:59 | sean-k-mooney | we could look at that or perhaps using --ignore-missing-imports | |
| 13:30:17 | stephenfin | We already have that configured via '[mypy] ignore_missing_imports = true' in setup.cfg | |
| 13:30:26 | stephenfin | apparently it doesn't work for things in typeshed | |
| 13:30:36 | sean-k-mooney | ya it is ignored for thing that were previously bundeled | |
| 13:30:41 | sean-k-mooney | https://mypy-lang.blogspot.com/2021/06/ | |
| 13:30:48 | sean-k-mooney | https://mypy-lang.blogspot.com/2021/06/mypy-0900-released.html | |
| 13:31:26 | sean-k-mooney | oh TypeGuard we can now do user defined narrowing | |
| 13:31:39 | amodi | lyarwood: https://review.opendev.org/c/openstack/whitebox-tempest-plugin/+/786979/20/whitebox_tempest_plugin/config.py#138 | |
| 13:34:32 | sean-k-mooney | stephenfin: do we want to consider adding mypy --install-types to the command | |
| 13:34:46 | sean-k-mooney | before we run mypy | |
| 13:34:51 | stephenfin | hmm, that would be a good short-term fix | |
| 13:35:04 | stephenfin | long-term though, I suspect these things might want to be tracked? | |
| 13:35:15 | sean-k-mooney | im not sure | |
| 13:35:20 | sean-k-mooney | they might want to be yes | |
| 13:35:43 | sean-k-mooney | but since its just for type checking im not sure | |
| 13:35:56 | stephenfin | fair point | |
| 13:35:59 | stephenfin | lemme try that | |
| 13:40:48 | sean-k-mooney | stephenfin: by the way we could use https://github.com/pre-commit/mirrors-mypy | |
| 13:41:09 | opendevreview | Rodrigo Barbieri proposed openstack/nova stable/victoria: Error anti-affinity violation on migrations https://review.opendev.org/c/openstack/nova/+/795542 | |
| 13:42:49 | stephenfin | 'mypy --install-types' is interactive | |
| 13:42:50 | stephenfin | :( | |
| 13:43:04 | stephenfin | so I have to pipe 'yes' into it | |
| 13:43:05 | stephenfin | lovely | |
| 13:43:35 | stephenfin | sean-k-mooney: https://github.com/python/mypy/issues/10600#issuecomment-857351152 | |
| 13:43:38 | stephenfin | see next comment | |
| 13:43:45 | stephenfin | I wouldn't recommend running --install-types as it currently works in CI, since in the worst case it can almost double the mypy runtime. It also produces noisy output, as mentioned above. | |
| 13:43:51 | stephenfin | </quote> | |
| 13:44:15 | stephenfin | so we can do it to unblock the gate, but it definitely shouldn't stay | |
| 13:44:26 | sean-k-mooney | stephenfin: ok i guess we can just run it once do a freeze before and after and see what it adds | |
| 13:44:31 | sean-k-mooney | if this happens again | |
| 13:44:37 | stephenfin | Successfully installed types-cryptography-0.1.1 types-enum34-0.1.5 types-ipaddress-0.1.2 types-paramiko-0.1.3 | |
| 13:44:58 | stephenfin | I guess not all of those are actually needed since we don't typecheck everything | |
| 13:45:21 | sean-k-mooney | not right now at least | |
| 13:45:35 | sean-k-mooney | but if we expand the checking they proably will be | |
| 13:46:13 | opendevreview | Stephen Finucane proposed openstack/nova master: requirements: Add types-paramiko https://review.opendev.org/c/openstack/nova/+/795533 | |
| 13:46:14 | opendevreview | Stephen Finucane proposed openstack/nova master: tox: Install mypy type stubs https://review.opendev.org/c/openstack/nova/+/795544 | |
| 13:46:24 | stephenfin | lyarwood, gibi: Alternate approach ^ | |
| 14:12:05 | spatel | sean-k-mooney morning! i am trying to delete one of vm which doesn't existing on compute machine (i can't see it on compute host). i am not able to delete it, any good way to clean up ? | |
| 14:13:14 | gibi | stephenfin: https://review.opendev.org/c/openstack/nova/+/795544 doesnt seem to work | |
| 14:17:41 | stephenfin | oh, well then | |
| 14:17:58 | stephenfin | so I guess you need to run it, let it build a cache, then run it again to install the missing deps | |
| 14:18:04 | stephenfin | so that won't work | |
| 14:18:10 | stephenfin | pity :-( | |
| 14:19:19 | opendevreview | Stephen Finucane proposed openstack/nova master: requirements: Add types-paramiko https://review.opendev.org/c/openstack/nova/+/795533 | |
| 14:19:46 | stephenfin | back to v2 we go | |
| 14:24:35 | sean-k-mooney | spatel: i think we have force delete option | |
| 14:24:56 | spatel | hmm --force ? | |
| 14:25:04 | sean-k-mooney | spatel: if the vm no longere exits on the compute node but is still in the nova db that is what i woudl try next | |
| 14:25:21 | spatel | perfect! let me try | |
| 14:25:40 | sean-k-mooney | well i was thinking of "nova force-delete <server uuid>" | |
| 14:25:45 | sean-k-mooney | let me check if its in osc | |
| 14:26:06 | sean-k-mooney | yes there is a --force | |
| 14:26:19 | sean-k-mooney | os openstack server delete --force <uuid> | |
| 14:27:00 | spatel | penstack server delete: error: unrecognized arguments: --force | |
| 14:27:12 | spatel | I am running victoria | |
| 14:28:27 | spatel | nova force-delete works! | |
| 14:29:44 | spatel | sean-k-mooney openstack server delete doesn't have --force option | |
| 14:30:00 | sean-k-mooney | it does in the latest release | |
| 14:30:34 | sean-k-mooney | spatel: stephenfin added it along with closing most of the other gaps recently | |
| 14:31:24 | spatel | may be available in current release but not in victoria | |
| 14:31:56 | sean-k-mooney | when i said recent i ment in the last month or so | |
| 14:32:08 | sean-k-mooney | nova force-delete does exactly the same thing | |
| 14:33:05 | opendevreview | sean mooney proposed openstack/osc-placement master: default to max version when no session https://review.opendev.org/c/openstack/osc-placement/+/794276 | |