Earlier  
Posted Nick Remark
#openstack-nova - 2021-08-03
16:27:23 stephenfin no other topics from me
16:27:44 gibi we should just close nova next week :D
16:28:14 gibi so if no other topic then that was it
16:28:17 gibi thanks for joining today
16:28:39 gibi #endmeeting
16:28:39 opendevmeet Meeting ended Tue Aug 3 16:28:39 2021 UTC. Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4)
16:28:39 opendevmeet Minutes: https://meetings.opendev.org/meetings/nova/2021/nova.2021-08-03-16.00.html
16:28:39 opendevmeet Minutes (text): https://meetings.opendev.org/meetings/nova/2021/nova.2021-08-03-16.00.txt
16:28:39 opendevmeet Log: https://meetings.opendev.org/meetings/nova/2021/nova.2021-08-03-16.00.log.html
16:28:50 bauzas gibi: thanks
16:28:56 gmann thanks gibi
16:29:01 gibi bauzas: thanks for running it last week
16:29:28 bauzas gibi: well, we had more people than last time I ran, so this was good :D
16:30:10 gibi :)
16:50:37 opendevreview Sylvain Bauza proposed openstack/nova master: Rename vgpu options to mdev https://review.opendev.org/c/openstack/nova/+/801607
16:50:37 opendevreview Sylvain Bauza proposed openstack/nova master: Provide the mdev class for every PCI device https://review.opendev.org/c/openstack/nova/+/802918
16:50:38 opendevreview Sylvain Bauza proposed openstack/nova master: Provide and use other RCs for mdevs if needed https://review.opendev.org/c/openstack/nova/+/803233
16:50:38 opendevreview Sylvain Bauza proposed openstack/nova master: Expose the mdev class https://review.opendev.org/c/openstack/nova/+/801743
16:50:39 opendevreview Sylvain Bauza proposed openstack/nova master: Change the admin-visible logs for mdev support https://review.opendev.org/c/openstack/nova/+/803378
16:50:39 opendevreview Sylvain Bauza proposed openstack/nova master: WIP: Cleanup GPU vs. mdev wording https://review.opendev.org/c/openstack/nova/+/803379
16:50:54 bauzas gibi: the last change is for you ^ ;)
16:51:29 gibi thank <3
16:51:29 gibi s
17:54:49 opendevreview Balazs Gibizer proposed openstack/nova master: Support interaface attach / detach with new resource request format https://review.opendev.org/c/openstack/nova/+/800089
17:56:07 opendevreview Balazs Gibizer proposed openstack/nova master: [func test] move unshelve test to the proper place https://review.opendev.org/c/openstack/nova/+/793621
17:56:08 opendevreview Balazs Gibizer proposed openstack/nova master: WIP support extended res req in heal port allocation https://review.opendev.org/c/openstack/nova/+/802060
20:03:07 opendevreview Merged openstack/nova stable/victoria: Move 'check-cherry-picks' test to gate, n-v check https://review.opendev.org/c/openstack/nova/+/797040
22:24:13 opendevreview Merged openstack/nova stable/ussuri: Move 'check-cherry-picks' test to gate, n-v check https://review.opendev.org/c/openstack/nova/+/797050
#openstack-nova - 2021-08-04
08:06:07 lyarwood FWIW https://review.opendev.org/c/openstack/nova/+/803322 has hit both https://bugs.launchpad.net/nova/+bug/1938021 and https://bugs.launchpad.net/nova/+bug/1936849 now, I've not had a chance to look at these really so if anyone has time this morning please take a look.
08:50:09 gibi lyarwood: ack, those bugs are on my radar too, but I have conflicting priorities :/
08:50:24 gibi I will try to look at them this week
08:54:54 bauzas lyarwood: i can try to take a look
08:55:35 bauzas in the meantime, can someone explain me the weirdo issue with mypy in https://review.opendev.org/c/openstack/nova/+/802918/7/nova/virt/libvirt/driver.py#511 ?
08:55:56 bauzas oh, because I used a lambda function
08:57:00 bauzas gibi: any idea how I could tell mypy to *not* ask for a type annotation for a lambda function ?
08:59:15 gibi bauzas: either you gave it a ty.Any annotation
08:59:16 gibi or
08:59:28 gibi there is some comment to silence mypy...
08:59:42 gibi # type: ignore
08:59:59 gibi bauzas: why don't you want to define the type? is it a complicated one?
09:00:25 bauzas gibi: I don't get why myty doesn't ask for an annotation here : https://github.com/openstack/nova/blob/master/nova/virt/libvirt/driver.py#L474
09:00:34 bauzas and why it asks for https://review.opendev.org/c/openstack/nova/+/802918/7/nova/virt/libvirt/driver.py#511
09:03:01 bauzas oh, the type annotation is for the new attribute
09:03:21 bauzas I guess because mypy can't statically guess it
09:03:35 bauzas as it uses a lambda function
09:03:50 gibi bauzas: you can check what mypy can guess reveal_type(<variable>)
09:04:23 gibi as far as I see local variables only resolved if the function is resolved by mypy
09:04:36 gibi so if the function has no annotation on the signature then mypy will not check the locals
09:05:32 bauzas which is the case for __init__
09:05:42 bauzas but
09:05:46 gibi would be interesting to see what type self._sysinfo_serial_func got from mypy
09:06:29 gibi but anyhow self.mdev_class_mapping is a Dict[str, str] isn't it?
09:06:30 bauzas let me run reveal_type on both mdev_class_mapping and pgpu_class_mapping
09:06:40 bauzas gibi: exact, annotation is simple
09:08:34 bauzas hum, very interesting
09:10:18 bauzas gibi: https://paste.opendev.org/show/807878/
09:10:46 bauzas if I ask reveal_type() this forces mypy to introspect local variables
09:11:06 bauzas so I guess we now need to add annotations every time
09:14:38 opendevreview Sylvain Bauza proposed openstack/nova master: Provide the mdev class for every PCI device https://review.opendev.org/c/openstack/nova/+/802918
09:14:39 opendevreview Sylvain Bauza proposed openstack/nova master: Provide and use other RCs for mdevs if needed https://review.opendev.org/c/openstack/nova/+/803233
09:14:39 opendevreview Sylvain Bauza proposed openstack/nova master: Expose the mdev class https://review.opendev.org/c/openstack/nova/+/801743
09:14:40 opendevreview Sylvain Bauza proposed openstack/nova master: WIP: Cleanup GPU vs. mdev wording https://review.opendev.org/c/openstack/nova/+/803379
09:21:34 gibi bauzas: I think that only means that mypy did not assign any type to either of those variables but mypy needed to use some type for mdev_class_mapping for something else later so mypy asked the typehint from you. When you added reveal_type you forced mypy to try to assign type to both variables hence is asked typehint for both from you
09:22:21 bauzas gibi: yup, that's what I found
09:22:30 bauzas static typing, my love
09:22:57 bauzas anyway, this is fixed in the last rev, I added an annotation
09:23:56 gibi ack
09:25:04 gibi I'm done with the re-review of the mdev series so far so good :)
09:25:24 bauzas see, I haven't grumbled about mypy
09:26:20 bauzas sure, it's important to tell that a variable using a collections.defaultdict is a dict :p
09:26:43 bauzas just in case people don't know :D
09:41:38 bauzas gibi: I replied to your (good) concern https://review.opendev.org/c/openstack/nova/+/803233/3//COMMIT_MSG#17
09:54:07 stephenfin kashyap: not an option, I'm afraid https://governance.openstack.org/tc/reference/tags/assert_follows-standard-deprecation.html
09:57:12 kashyap Yeah, I know "standards"
10:32:27 sean-k-mooney kashyap: well in this case its openstacks standard
10:32:47 sean-k-mooney which nova does follow
10:36:46 sean-k-mooney kashyap: for what its worth if qemu upstream dont remove it im not really in a rush to remove it form nova. deprecating floppy usage sure, but it does not cost us much if anything to keep it and distros can still drop support even if we support it upstream
10:39:05 kashyap sean-k-mooney: Hey; I see what you mean, but as noted it is more of a liability than anything at this point
10:39:17 kashyap (Given past CVEs. Upstream QEMU too, it's discouraged)
10:40:05 kashyap (Downstream distros of OpenStack can also simply declare it out of scope / deprecated.)
10:55:07 opendevreview Merged openstack/nova master: zuul: Increase GLANCE_LIMIT_IMAGE_SIZE_TOTAL for nova-lvm https://review.opendev.org/c/openstack/nova/+/803322
11:00:02 lyarwood \o/
11:37:28 opendevreview Lee Yarwood proposed openstack/nova master: libvirt: Handle silent failures to extend volume within os-brick https://review.opendev.org/c/openstack/nova/+/801714
11:37:55 opendevreview Lee Yarwood proposed openstack/nova master: Add functional test for bug 1937375 https://review.opendev.org/c/openstack/nova/+/802011
11:37:56 opendevreview Lee Yarwood proposed openstack/nova master: compute: Avoid duplicate BDMs during reserve_block_device_name https://review.opendev.org/c/openstack/nova/+/801990
11:37:56 opendevreview Lee Yarwood proposed openstack/nova master: fup: Move _wait_for_volume_attach into InstanceHelperMixin https://review.opendev.org/c/openstack/nova/+/802623
11:38:26 opendevreview Lee Yarwood proposed openstack/nova master: Add regression test for bug 1938326 https://review.opendev.org/c/openstack/nova/+/802801
11:38:26 opendevreview Lee Yarwood proposed openstack/nova master: compute: Do not mark disabled but down services as in maintenance https://review.opendev.org/c/openstack/nova/+/802317
11:38:46 lyarwood gibi / bauzas ; now the gate is fixed reviews on ^ would be appreciated this week if you have time
11:45:26 gibi lyarwood: on it
12:01:03 gibi lyarwood: after your lunch, is this extended race scenario possible? https://review.opendev.org/c/openstack/nova/+/801990/6/nova/compute/manager.py#6985
12:12:11 lyarwood gibi: yeah that's also possible but I wonder if we want to treat it as a separate fix?
12:13:45 lyarwood actually thinking about it, is it an issue if we have racing requests against the same volume
12:14:58 lyarwood multiattach races would be caught by c-api later
12:15:00 gibi lyarwood: I would treat it as a separate fix. whay you proposed is a valid fix for a scenario reported int he bug
12:15:59 gibi hm, I guess we have the instance.uuid lock as a pattern, we have that for most of the operations
12:16:15 gibi but here a volume.id lock would be better
12:16:26 gibi I'm not sure we even need the instance.uuid lock here
12:16:32 gibi but could be
12:27:20 stephenfin bauzas: Can we trade reviews for your generic mdev series and my DB series? I'm really eager to close out as much as I can this week, especially with a few people on PTO next week

Earlier   Later