Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2018-03-30
15:30:19 fried_bunny sorry guys, was afk for a few. Catching up.
15:31:05 fried_bunny mriedem: Ack
15:31:21 fried_bunny leakypipes: Yes, was planning to dig into that today.
15:33:45 leakypipes fried_bunny: cool, np. just checking to see if you had
15:33:57 fried_bunny yuh, thanks.
15:34:20 leakypipes cdent: understatement of the year in your placement update: "There's also been a _lot_ of discussion around NUMA handling"
15:34:30 fried_bunny leakypipes: I've slowed myself way down by picking up lots of extra reviews - probably too many, need to scale back some.
15:34:46 leakypipes fried_bunny: learn, you will, young paduan.
15:37:36 openstackgerrit Matt Riedemann proposed openstack/nova master: Add "bind_ports_to_host" neutron API method https://review.openstack.org/523604
15:37:37 openstackgerrit Matt Riedemann proposed openstack/nova master: WIP: libvirt: use dest host vif migrate details for live migration https://review.openstack.org/551370
15:37:37 openstackgerrit Matt Riedemann proposed openstack/nova master: Add VIFMigrateData object for live migration https://review.openstack.org/515423
15:37:38 openstackgerrit Matt Riedemann proposed openstack/nova master: Add "activate_port_binding" neutron API method https://review.openstack.org/555947
15:37:38 openstackgerrit Matt Riedemann proposed openstack/nova master: Add "delete_port_binding" network API method https://review.openstack.org/552170
15:37:39 openstackgerrit Matt Riedemann proposed openstack/nova master: Implement migrate_instance_start method for neutron https://review.openstack.org/556334
15:37:39 openstackgerrit Matt Riedemann proposed openstack/nova master: Delete port bindings in setup_networks_on_host if teardown=True https://review.openstack.org/556333
15:37:40 openstackgerrit Matt Riedemann proposed openstack/nova master: WIP: Port binding based on events during live migration https://review.openstack.org/434870
15:37:40 openstackgerrit Matt Riedemann proposed openstack/nova master: WIP: compute: use port binding extended API during live migration https://review.openstack.org/551371
15:37:41 openstackgerrit Matt Riedemann proposed openstack/nova master: conductor: use port binding extended API in during live migrate https://review.openstack.org/522537
15:39:54 openstackgerrit Merged openstack/nova-specs master: Update the vGPU spec https://review.openstack.org/557912
16:15:55 openstackgerrit Arvind Nadendla proposed openstack/nova master: Update ImageMetaProp object to expose traits https://review.openstack.org/557795
16:16:35 openstackgerrit Ed Leafe proposed openstack/nova master: Add handlers for consumer_generation https://review.openstack.org/557959
16:16:35 openstackgerrit Ed Leafe proposed openstack/nova master: Add the 'generation' column to the consumers table https://review.openstack.org/557958
16:17:06 figleaf leakypipes: fried_bunny: cdent: ^^ First crack at consumer generations. More to come
16:17:39 fried_bunny figleaf: ack
16:17:52 leakypipes figleaf: nice, thank you sir.
16:18:20 cdent figleaf: I'm going to ignore that until next week please and thank you
16:19:06 fried_bunny claudiub|2: Still around?
16:23:16 figleaf cdent: ignore away
16:23:19 figleaf :)
16:26:58 claudiub|2 fried_bunny: i am now
16:28:04 cdent That's it for me today. Happy Easter if that's your thing, happy weekend otherwise.
16:28:29 claudiub|2 the autospec stuff?
16:28:37 claudiub|2 i''l take a look
16:48:43 openstackgerrit Ed Leafe proposed openstack/nova master: Add handlers for consumer_generation https://review.openstack.org/557959
16:56:24 fried_bunny claudiub|2: Thank you sir.
16:57:55 fried_bunny claudiub|2: The failures are really weird. The leading '<' is stripped from the mock repr in the KeyError. Not sure if that's significant.
16:58:18 openstackgerrit Roman Dobosz proposed openstack/nova master: Added weight for aggregate soft (anti) affinity. https://review.openstack.org/556762
17:12:48 openstackgerrit Merged openstack/nova stable/pike: Only attempt a rebuild claim for an evacuation to a new host https://review.openstack.org/550555
17:12:56 openstackgerrit Merged openstack/nova master: Fix comments in get_all_with_shared() https://review.openstack.org/533195
17:13:03 openstackgerrit Merged openstack/nova stable/pike: Add --by-service to discover_hosts https://review.openstack.org/554603
17:13:14 openstackgerrit Merged openstack/nova master: Add AggregateList.get_by_metadata() query method https://review.openstack.org/544728
17:16:32 fried_bunny claudiub|2, edmondsw: It's specifically this part that breaks us:
17:16:33 fried_bunny When I comment out *just* that fixture, we're fine.
17:16:33 fried_bunny https://review.openstack.org/#/c/447505/44/nova/test.py
17:16:44 fried_bunny Which isn't surprising; that's the only part that could have possibly affected us OOT.
17:16:50 fried_bunny So - what IS that thing?
17:20:16 claudiub|2 you mean what is that fixture?
17:20:19 fried_bunny yeah
17:20:36 claudiub|2 it adds the autospec argument to mock.Mock and mock.MagicMock
17:21:16 claudiub|2 which is even more strange, because I don't see how it could have affected this
17:22:28 fried_bunny claudiub|2: Well, we're inheriting nova.test.TestCase, so that's how we're picking up the change. That much is clear.
17:22:35 claudiub|2 since I don't see autospecs being used anywhere
17:23:23 fried_bunny claudiub|2: By "it adds the autospec argument" do you mean that it just makes it available, not that it actually sets it to anything?
17:24:28 claudiub|2 yep
17:24:41 claudiub|2 if autospec is not specifically given, it is noop
17:24:51 claudiub|2 question, do you split by underscore somewhere?
17:25:00 fried_bunny oh, I'm sure we must
17:25:15 claudiub|2 oh
17:25:18 claudiub|2 that might be it
17:25:45 claudiub|2 yeah, I think I know what's happening
17:25:46 fried_bunny ...not seeing it though.
17:25:58 fried_bunny okay, do tell. Cause I'm stumped. Mystified. Stupefied.
17:27:48 claudiub|2 so, you're using a lot of thing.uuid, or thing.name and so on. so, if that property is not set on the mock, something like this is printed: str(mock_thing.foo) -> <_AutospecMock name='mock.foo' id='139796718268752'>
17:28:23 claudiub|2 the mock class with autospec is called _AutospecMock
17:28:57 claudiub|2 ... and since you're splitting the wrappers by _, it's split in 2 places
17:29:21 claudiub|2 because your wrappers are called wrapper_<_AutospecMock name=...>
17:29:32 fried_bunny claudiub|2: What if I don't see any splits anywhere?
17:29:47 fried_bunny ...by underscore
17:30:07 claudiub|2 you sure? I think I saw one
17:30:41 claudiub|2 yeh
17:30:43 claudiub|2 there is
17:31:27 claudiub|2 pypowervm/utils/transaction.py L 744
17:31:37 claudiub|2 you're doing a rsplit
17:31:47 claudiub|2 ehm, why rsplit? and not lsplit?
17:33:02 claudiub|2 ehm, rsplit is just split. :))
17:33:08 fried_bunny oy vay, my search was broken. I see it now.
17:33:13 fried_bunny claudiub|2: I don't think so...
17:33:51 fried_bunny Out[3]: ['one', 'two_three']
17:33:51 fried_bunny In [3]: 'one_two_three'.split('_', 1)
17:33:51 fried_bunny Out[2]: ['one_two', 'three']
17:33:51 fried_bunny In [2]: 'one_two_three'.rsplit('_', 1)
17:33:51 fried_bunny claudiub|2:
17:34:00 openstackgerrit Merged openstack/nova master: Add require_tenant_aggregate request filter https://review.openstack.org/545002
17:34:05 fried_bunny claudiub|2: It matters when limiting the split
17:34:09 openstackgerrit Merged openstack/nova master: Documentation for tenant isolation with placement https://review.openstack.org/557490
17:34:42 claudiub|2 right, anyways, the thing is, the underscore matters apparently
17:34:57 claudiub|2 I renamed _AutospecMock to AutospecMock and it worked
17:34:58 fried_bunny claudiub|2: And it's because IRL the label part can have underscores, and we want to preserve them.
17:35:34 fried_bunny claudiub|2: ...and the UUID doesn't have underscores. Unless it's a str(mock)ification, apparently :P
17:36:28 claudiub|2 indeed. :)
17:36:41 claudiub|2 well, ok, there are 2 solutions to the problem
17:37:06 claudiub|2 one is a bit of a PINTA, so I'll just submit the easier one, but it'll require a new oslotest release
17:37:35 fried_bunny claudiub|2: The other one is for us to go set our UUIDs in our mocks?
17:37:42 fried_bunny rather than letting them mock themselves
17:38:10 claudiub|2 the easier solution is to rename _AutospecMock and _AutospecMagicMock in oslotest, basically remove that _. I've put that _ there so it would discourage people for using them directly, but meh.
17:38:23 claudiub|2 and the other one is as you say
17:40:50 fried_bunny claudiub|2: Okay, please add me to that review.
17:40:58 fried_bunny ...and I'll go see if I can find those UUIDs...
17:43:02 claudiub|2 fried_bunny: added: https://review.openstack.org/#/c/557974/
17:43:23 claudiub|2 can you run the UT with this patch applied?

Earlier   Later