Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2025-02-21
03:48:58 masahito s/no input in show/no input None in show/
03:48:58 sean-k-mooney sure but im trying to decied if that matters
03:49:46 sean-k-mooney if we dont have a request spec then it cant be pinned
03:49:52 masahito i see.
03:50:41 sean-k-mooney im not sayign you should change this
03:51:00 sean-k-mooney im trying to undersand the desing chooices that lead to this current patch
03:51:29 sean-k-mooney "if provided_az is False:" is very specific and odd
03:51:30 masahito thanks for the follow up. my reply is one previous your message.
03:51:39 sean-k-mooney you are using False as a sential
03:51:51 sean-k-mooney but it might be better to actuly create a new sentail object instead
03:51:55 sean-k-mooney and use that instead
03:52:09 sean-k-mooney so its clear what its role is
03:52:40 masahito like a empty request_spec object instead of False?
03:52:56 sean-k-mooney no
03:53:07 sean-k-mooney so at teh module level define sentinel = object()
03:53:40 sean-k-mooney then do req_specs_dict.get(server.uuid, sentinel) here
03:53:42 sean-k-mooney https://review.opendev.org/c/openstack/nova/+/939658/6/nova/api/openstack/compute/views/servers.py#539
03:54:03 sean-k-mooney and if provided_az is sentinel:
03:54:05 sean-k-mooney https://review.opendev.org/c/openstack/nova/+/939658/6/nova/api/openstack/compute/views/servers.py#222
03:55:18 sean-k-mooney rather then using the singleton False object create your own
03:55:33 sean-k-mooney you can give it a descriptive name
03:55:55 sean-k-mooney AZ_NOT_IN_REQUEST_SEPC = obejct()
03:55:56 masahito I may get your point. at the first if-condition checks the provided_az is the sentinel object or not.
03:56:08 sean-k-mooney right
03:56:16 sean-k-mooney basiclly https://python-patterns.guide/python/sentinel-object/#sentinel-objects
03:57:54 sean-k-mooney ill leave a comment to this effect and also +1
03:58:17 sean-k-mooney letss see if others care one way or the other
03:59:23 masahito Let me update to use the sentinel. I don't think it difficult change.
03:59:33 sean-k-mooney sure
04:07:18 opendevreview Masahito Muroi proposed openstack/nova master: Use dict object for request_specs_dict in the _list_view https://review.opendev.org/c/openstack/nova/+/939658
04:10:35 masahito updated. thanks
04:10:54 sean-k-mooney i have one nit in the release note but im +2
04:38:38 sean-k-mooney melwitt: im not sure if you wrote that patch or not but im actully got to head to sleep now, enjoy whats left of your evnening
04:39:19 melwitt sean-k-mooney: no not yet. gnight o/
04:57:43 sshmlbrg sean-k-mooney: o/
05:30:27 opendevreview sean mooney proposed openstack/nova master: only show standard image propeties in server show. https://review.opendev.org/c/openstack/nova/+/942413
05:30:44 sean-k-mooney gibi: ^ that shoudl fix
05:30:49 sean-k-mooney https://launchpad.net/bugs/2098384
05:33:05 sean-k-mooney well maybe not
05:33:15 sean-k-mooney this is ligtly diffent damb it
05:35:50 sean-k-mooney well that fixes a bug but it might only be a partial fix
05:55:42 opendevreview Merged openstack/nova-specs master: Extend image-properties-in-server-show spec with rebuild https://review.opendev.org/c/openstack/nova-specs/+/940360
06:59:21 mikal sean-k-mooney / melwitt: there is this WIP tempest patch for v2.98 -- https://review.opendev.org/c/openstack/tempest/+/939375
06:59:57 mikal I am a little surprised its not in merge conflict now though.
07:12:21 mikal Oh wait, that's the _other_ _other_ competitor for v2.98? Is there any existing attempt at testing for v2.98? If it changed the schema how did it merge without any tempest tests?
07:12:51 mikal I am confused enough that I am going to stop and think about this some more.
07:18:42 mikal I cannot find a proposed tempest change for the code which landed as v2.98.
07:42:31 opendevreview Johannes Kulik proposed openstack/nova master: libvirt: fix maxphysaddr passthrough dom parsing https://review.opendev.org/c/openstack/nova/+/942371
08:36:07 melwitt sean-k-mooney: I saw your comments on the image properties patch and your fix patch. I assume that means I should hold off on trying any tempest schema patch for 2.98 for the moment if there is some bug in the API response?
08:39:21 ratailor sean-k-mooney, you around ?
08:40:27 sshmlbrg ratailor: he's sleeping
08:41:33 ratailor sshmlbrg, ack. Thanks!
10:23:01 opendevreview Rajesh Tailor proposed openstack/nova master: only show standard image properties in server show. https://review.opendev.org/c/openstack/nova/+/942413
11:11:48 sean-k-mooney melwitt: it does not change teh schema becuase in the schema its just a dict of sting
11:41:21 opendevreview Vasyl Saienko proposed openstack/nova-specs master: Expose vlan trunking in metadata/configdrive https://review.opendev.org/c/openstack/nova-specs/+/471815
12:04:15 nisha hi nova folks, please review the one line bug fix for port after live migration if you have time. thank you! https://review.opendev.org/c/openstack/nova/+/936338
12:04:42 sean-k-mooney nisha: that not a trivile change
12:05:05 sean-k-mooney i skimmed it when you first pushed it and i did have tiem to fully load all the context
12:06:48 sean-k-mooney there are some filed that should be diffent between the souce and dest
12:07:39 sean-k-mooney for example the pci adress of an sriov nic.
12:08:25 sean-k-mooney so this need careful review to ensure its actully correct
12:09:16 sean-k-mooney the binding provile is an admin only filed that is ment to be used by nova to pass info to the neutron network backend
12:09:39 sean-k-mooney its not a freform field that operators or end user are allowed to put arbitary keys in
12:10:38 nisha ahh I see. thank you for taking a look at it already.
12:11:16 sean-k-mooney what filed were being lost in your case
12:11:59 sean-k-mooney do you have a specific case that is broken
12:12:48 sean-k-mooney {"os_vif_delegation": true} was not ment to be the only content by the way it was ment to be added to the prociel and remvoe dlater
12:14:23 sean-k-mooney the bug doe snot say it just say "openstack port set $port_id --binding-profile aaa=bbb --binding-profile ccc=ddd"
12:14:29 sean-k-mooney which is not supproted
12:15:10 sean-k-mooney there is only one filed that can be set by a human in the bidnign profile ant that trusted_vf
12:15:32 sean-k-mooney and removing that has been on our todo list forever
12:21:46 nisha in our case, the bgp network settings are set to the port in the binding profile and after live migration it is completely overwritten by os_vif_delegation
12:27:51 sean-k-mooney that doesn not sound like a supprote setup. can you point to the neutron pluging and confirm what virt driver your using
12:29:12 sean-k-mooney in the case of use with nova the content of the binding profile is exlicively resrevd for novas use, for zun it reserved for zun for standalone ironic it cna use it but it may not modify it when used via nova
12:29:53 sean-k-mooney there is a bug on the nova side but its not ok for neturon to modify the content of that filed
12:30:18 sean-k-mooney so if the ml2 dirver is modifying its incorrect
12:35:50 nisha yes, ml2 is the neutron core plugin and libvirt driver
12:36:06 sean-k-mooney that does not answer my question
12:36:13 sean-k-mooney which ml2 driver?
12:36:17 sean-k-mooney there are many
12:49:05 cardoe vsaienko: so fwiw, your trunk details patch https://review.opendev.org/c/openstack/nova/+/941227 with my suggested change works for us on real hardware
12:54:27 opendevreview Vasyl Saienko proposed openstack/nova-specs master: Expose vlan trunking in metadata/configdrive https://review.opendev.org/c/openstack/nova-specs/+/471815
12:55:37 vsaienko cordoe: thanks, nova team asked to create a spec https://review.opendev.org/c/openstack/nova-specs/+/471815, and I've discovered that in my patch grenade job is failing because of issues with Instance object format not recognized correctly by older nova-compute.
13:05:40 opendevreview Luis Enrique Morales Mendez proposed openstack/nova master: Add 'LOG' to archive_deleted_rows function https://review.opendev.org/c/openstack/nova/+/942441
13:06:59 cardoe Looking to land https://review.opendev.org/c/openstack/nova/+/942019 and then create some backports to 2024.2 and 2024.1 without that change those versions just log the SDK Result Object at <0xDEADBEEF> as the error message.
13:12:47 nisha sean-k-mooney: we use custom driver which is a combination of srv6 and l2isolate type driver using linuxbridge based agent, i suspect this is responsible to overwrite the binding_profile. i understand the current situation on nova's side. thank you for sharing the details.
15:57:53 ratailor sean-k-mooney, you around ?
21:01:56 opendevreview melanie witt proposed openstack/nova master: DNM test microversion 2.99 https://review.opendev.org/c/openstack/nova/+/942493
21:11:25 mikal melwitt / sean-k-mooney: https://review.opendev.org/c/openstack/tempest/+/942495 is an attempt at the schema changes in tempest for v2.98. I am fairly confident it is wrong.
21:18:21 melwitt mikal: I also uploaded one a few minutes ago heh. I'm running a DNM patch at the end of the series to see if it works
21:18:40 opendevreview Michael Still proposed openstack/nova master: libvirt: direct SPICE console object changes https://review.opendev.org/c/openstack/nova/+/926876
21:18:40 opendevreview Michael Still proposed openstack/nova master: libvirt: direct SPICE console database changes https://review.opendev.org/c/openstack/nova/+/926877
21:18:41 opendevreview Michael Still proposed openstack/nova master: libvirt: allow direct SPICE connections to qemu https://review.opendev.org/c/openstack/nova/+/924844
21:18:58 mikal melwitt: oh I saw your nova one, did you do a tempest one too?
21:19:10 melwitt aaand it just got orphaned I think
21:19:14 melwitt haha, yes I did
21:19:17 mikal melwitt: I might have just stomped on yours then...
21:19:28 mikal Ok, I'll revert mine changes to the nova series. Sorry.
21:19:36 melwitt no I think it's ok
21:19:58 mikal Well the series now depends on my lame attempt at a v2.98 schema in tempest

Earlier   Later