| *** tetsuro has joined #openstack-placement | 00:34 | |
| *** takashin has joined #openstack-placement | 02:26 | |
| *** bhagyashris has joined #openstack-placement | 02:51 | |
| *** dklyle has joined #openstack-placement | 03:23 | |
| *** belmoreira has quit IRC | 04:10 | |
| *** zzzeek has quit IRC | 04:49 | |
| *** zzzeek has joined #openstack-placement | 04:52 | |
| *** irclogbot_2 has quit IRC | 04:55 | |
| *** e0ne has joined #openstack-placement | 06:25 | |
| *** e0ne has quit IRC | 06:26 | |
| *** belmoreira has joined #openstack-placement | 07:16 | |
| *** e0ne has joined #openstack-placement | 07:23 | |
| *** takashin has left #openstack-placement | 08:01 | |
| *** e0ne has quit IRC | 08:13 | |
| *** helenafm has joined #openstack-placement | 08:21 | |
| *** tssurya has joined #openstack-placement | 08:31 | |
| *** evrardjp_ is now known as evrardjp | 08:31 | |
| *** ttsiouts has joined #openstack-placement | 08:57 | |
| *** ttsiouts has quit IRC | 09:08 | |
| *** ttsiouts has joined #openstack-placement | 09:08 | |
| *** ttsiouts has quit IRC | 09:13 | |
| *** ttsiouts has joined #openstack-placement | 09:25 | |
| *** ttsiouts has quit IRC | 09:38 | |
| *** ttsiouts has joined #openstack-placement | 09:38 | |
| *** tetsuro has quit IRC | 09:41 | |
| *** e0ne has joined #openstack-placement | 09:53 | |
| *** bhagyashris has quit IRC | 09:55 | |
| *** ttsiouts has quit IRC | 09:56 | |
| *** ttsiouts has joined #openstack-placement | 09:58 | |
| *** cdent has joined #openstack-placement | 10:06 | |
| *** ttsiouts has quit IRC | 10:40 | |
| *** ttsiouts has joined #openstack-placement | 10:41 | |
| *** ttsiouts_ has joined #openstack-placement | 10:42 | |
| *** ttsiouts has quit IRC | 10:45 | |
| *** ttsiouts_ has quit IRC | 11:57 | |
| openstackgerrit | Chris Dent proposed openstack/placement master: Use one ConfigOpts in placement-status https://review.openstack.org/633595 | 12:52 |
|---|---|---|
| *** ttsiouts has joined #openstack-placement | 13:11 | |
| *** ttsiouts has quit IRC | 13:18 | |
| *** ttsiouts has joined #openstack-placement | 13:19 | |
| *** rubasov has quit IRC | 13:33 | |
| *** ttsiouts has quit IRC | 14:04 | |
| *** ttsiouts has joined #openstack-placement | 14:04 | |
| *** ttsiouts has quit IRC | 14:09 | |
| *** ttsiouts has joined #openstack-placement | 14:09 | |
| *** mriedem has joined #openstack-placement | 14:25 | |
| *** mriedem has left #openstack-placement | 14:40 | |
| *** mriedem has joined #openstack-placement | 14:40 | |
| *** rubasov has joined #openstack-placement | 15:09 | |
| mriedem | can anyone remind me why when i tox -r -e functional-py35 in the placement repo it doesn't pull in os-resource-classes? | 15:18 |
| mriedem | last time i hit this i had to source the venv and manually install that library to get tests to work | 15:18 |
| mriedem | https://bugs.launchpad.net/nova/+bug/1816625 | 15:21 |
| openstack | Launchpad bug 1816625 in OpenStack Compute (nova) "tox -r fails to install os-resource-classes in extracted placement repo when running tests" [Undecided,New] | 15:21 |
| cdent | mriedem: hmmm. that's odd. I've not had that issue. | 15:30 |
| openstackgerrit | Matt Riedemann proposed openstack/placement master: Downgrade os-traits/os-resource-classes sync log level to DEBUG https://review.openstack.org/637892 | 15:32 |
| mriedem | cdent: what version of pip do you have? | 15:32 |
| cdent | in the venv I end up with 19.x. I'm using tox 3.7.0 | 15:34 |
| mriedem | yeah pip 19.0.2 in the venv | 15:35 |
| mriedem | tox 3.1.1 but i hit this a few weeks ago as well on older tox | 15:35 |
| mriedem | it's almost like it thinks those libraries are already installed and doesn't try to install them again | 15:35 |
| efried | fwiw, I just did functional-py36 in a clean repo and it pulled okay. | 15:35 |
| cdent | mriedem: anything useful in .tox/py35/log ? | 15:36 |
| mriedem | log is empty | 15:37 |
| mriedem | i still have a 'shared' direction under .tox/ - should that be gone? | 15:38 |
| cdent | yes, but I would have thought that would clean itself up | 15:38 |
| cdent | maybe just wipe out .tox and see if it is still there from scratch? | 15:38 |
| mriedem | doing that now | 15:39 |
| cdent | and then we can get stephenfin to explain everything | 15:39 |
| stephenfin | I didn't do it | 15:39 |
| mriedem | hit it again | 15:40 |
| mriedem | the import error | 15:40 |
| cdent | wacky | 15:40 |
| mriedem | shared is gone, and tox log is still empty | 15:40 |
| cdent | any chance you are overriding the upper constraints env var, or have out of date requiements? | 15:41 |
| mriedem | os-resource-classes is in requirements.txt | 15:41 |
| cdent | but not listed in installed... wacky | 15:41 |
| cdent | try upgrading tox? | 15:42 |
| mriedem | i doubt that will do anything b/c i hit this a few weeks ago on older tox and upgraded to 3.1.1 last week b/c of the minimum version bump on tox for nova | 15:45 |
| mriedem | just going to blast the repo and re-clone | 15:46 |
| cdent | mriedem: good luck | 15:47 |
| mriedem | well f me sideways | 15:48 |
| cdent | if that doesn't work, I'd try upgrading tox because efried and I have both made changes in the not too distant past that think in terms of modern tox | 15:48 |
| cdent | wuh? | 15:48 |
| mriedem | now it's fine | 15:48 |
| cdent | after the wipe? | 15:48 |
| mriedem | yeah | 15:49 |
| cdent | sigh | 15:49 |
| mriedem | oh well | 15:49 |
| cdent | this just confirms my stance of hating everything | 15:49 |
| efried | cdent: We're still importing placement fixtures from nova? | 15:53 |
| cdent | it imports the nova fixture which is a subclass of the placement provided one | 15:54 |
| cdent | because the nova one wanted additional chrome | 15:54 |
| cdent | (things like monkey patching the report client) | 15:54 |
| efried | yeah. I'm going to get rid of that now. | 15:54 |
| cdent | the monkey patching? | 15:54 |
| efried | (he says, determinedly) | 15:54 |
| efried | yeah. I remember thinking it wouldn't be all that hard, but if I didn't do it at the time, maybe it was because it was gonna be harder than I thought? | 15:55 |
| efried | do you remember anything that's going to help me do less brick-wall-banging here? | 15:55 |
| cdent | All I recall was that we didn't do it at the time because it wasn't an immediate requirement | 15:57 |
| efried | ahem, placement doesn't use its own PlacementFixture? | 15:57 |
| efried | ...it uses PlacementDirect | 15:58 |
| efried | The gears are starting to turn | 15:58 |
| efried | I think I was advocating getting rid of PlacementFixture in favor of PlacementDirect itself. | 15:58 |
| efried | cdent: that still keeps the dep from nova to placement, but it would at least consolidate (and let us get rid of those horrendous monkey patches) | 15:59 |
| *** e0ne has quit IRC | 16:00 | |
| efried | ...except I can't even see where it uses PlacementDirect. | 16:00 |
| cdent | you're confusing me | 16:01 |
| cdent | PlacementFixture is for external-to-placement tests, mostly | 16:01 |
| cdent | PlacementDirect is for an imaginary use case that never came to pass | 16:01 |
| efried | cdent: so what does placement itself use for its own internal functional tests (that aren't under db/)? | 16:02 |
| *** e0ne has joined #openstack-placement | 16:03 | |
| efried | like, how tf do the gabbits work? | 16:03 |
| cdent | the same way they always have: you pass a wsgi application to gabbi and it "takes care of it" | 16:03 |
| cdent | (looks like PlacementDirect is used in functional/test_verify_policy.py (as a convenient way to establish the wsgi intercept) | 16:04 |
| cdent | what gabbi does under the covers is establish the wsgi-intercept, but it does it per test suite, not per test | 16:05 |
| cdent | are the dots connecting yet efried or something else? | 16:06 |
| efried | cdent: Still gaps in brain. | 16:07 |
| efried | cdent: I remember thinking quite strongly at some point that we didn't need the fake_<HTTP op> methods in the nova fixture. | 16:07 |
| cdent | right because an adapter could be used instead | 16:08 |
| efried | ...but it is. | 16:08 |
| efried | I mean, the fake methods are passing through to the adapter. | 16:08 |
| cdent | the idea was to wrap the adapter in an adapter, as I recall | 16:09 |
| cdent | but I don't remember the details that much | 16:09 |
| cdent | I suspect there's code that can be borrowed from PlacementDirect | 16:11 |
| *** jaypipes has quit IRC | 16:36 | |
| *** jaypipes has joined #openstack-placement | 16:36 | |
| *** ttsiouts has quit IRC | 16:56 | |
| *** ttsiouts has joined #openstack-placement | 16:56 | |
| *** tssurya has quit IRC | 17:01 | |
| *** ttsiouts has quit IRC | 17:01 | |
| *** helenafm has quit IRC | 17:01 | |
| *** e0ne has quit IRC | 17:35 | |
| *** rubasov has quit IRC | 17:53 | |
| *** tssurya has joined #openstack-placement | 18:28 | |
| *** e0ne has joined #openstack-placement | 18:41 | |
| *** e0ne has quit IRC | 18:42 | |
| efried | cdent: Help! | 18:48 |
| cdent | eh? | 18:48 |
| efried | cdent: https://review.openstack.org/637959 | 18:48 |
| efried | cdent: I am having trouble with nova.tests.functional.regressions.test_bug_1679750 | 18:49 |
| * cdent reads | 18:49 | |
| cdent | yeah that test has been difficult two times in the past: | 18:51 |
| cdent | when I changed it to use the PlacementFixture | 18:51 |
| efried | I figured | 18:51 |
| cdent | and when mriedem (I think) wrote it in the first place | 18:51 |
| cdent | because of that bit about "conditionally" | 18:51 |
| efried | yeah | 18:51 |
| cdent | What's going wrong for you? | 18:51 |
| efried | well, when I leave it as is, all the placement calls from the compute service seem to work, but then the GET /a_c call (which comes from the scheduler service, yah?) goes 401. | 18:52 |
| * cdent tries it | 18:53 | |
| *** e0ne has joined #openstack-placement | 18:54 | |
| efried | which I initially thought is probably because the placement fixture isn't in play while the scheduler service is being started. | 18:54 |
| efried | cdent: Lemme summarize what I changed about the fixture, cause that may help. | 18:54 |
| efried | Before, it was mocking out the report client's get/put/post/delete methods, so the mock interception was happening late. | 18:55 |
| efried | Now I'm mocking out the report client's adapter arg at init time, so the fixture needs to be in place by the time the report client is created. | 18:55 |
| efried | if it's not, it's gonna try to use its own adapter. | 18:56 |
| efried | cdent: Sorry, before you get started on that test case, you have to s/\.body/.json()/ in three places. | 18:56 |
| efried | because I "fixed" the API | 18:57 |
| cdent | yeah, figured that out from reading the diff | 18:57 |
| efried | cdent: When I wrap the scheduler service start in the fixture cm, I get this: | 19:00 |
| efried | SAWarning: Evaluating non-mapped column expression 'id' onto ORM instances; this is a deprecated use case. Please make use of the actual mapped columns in ORM-evaluated UPDATE / DELETE expressions. | 19:00 |
| efried | (and of course the purpose of the "intermittent fixturing" test case is nullified) | 19:01 |
| cdent | efried: sorry I'm still getting the test to run: im_func is a python2ism. Use __func__ | 19:01 |
| efried | ah, mahbad | 19:01 |
| cdent | hmm, but that seems to have made no difference, but is still an issue: | 19:02 |
| efried | at least I got __defaults__ right | 19:02 |
| * efried creates py3 env... | 19:03 | |
| cdent | I can't get past it trying to create the ReportClient mock | 19:11 |
| efried | hm, yeah, in ipython I can't get __init__'s __func__ out. | 19:13 |
| efried | in fact __init__ doesn't seem to have any properties. | 19:14 |
| efried | cdent: do you have a py2 env? | 19:15 |
| cdent | not yet, but heading that way | 19:15 |
| cdent | efried: I'm being called to dinner, but will come back afters and continue digging. That should give you enough time to figure it all out. | 19:19 |
| efried | cdent: Thanks. | 19:19 |
| *** e0ne has quit IRC | 19:39 | |
| *** e0ne has joined #openstack-placement | 19:39 | |
| cdent | efried: I'm back | 19:44 |
| *** e0ne has quit IRC | 19:45 | |
| efried | cdent: I decided I needed to give up for now. Though I haven't figured out how I'm going to put it down. | 19:50 |
| efried | cdent: I can restore the old JIT-mock behavior by continuing to mock the get/put/post/delete methods through the fake adapter in a slightly different way, but that's not a lot of gain. | 19:51 |
| cdent | maybe nothing? what's there works, yeah? | 19:52 |
| cdent | (there == in master) | 19:52 |
| cdent | you can pick it up wholesale at any point you like in the future | 19:52 |
| efried | Well, it works except that it's making the behavior enough different than the real behavior to be problematic (viz. the issue that prompted Adam to write https://review.openstack.org/#/c/637657/) and that reduces confidence that the real code is doing the right thing just because the tests work. | 19:53 |
| efried | and also, it shouldn't be freakin necessary. | 19:54 |
| efried | so it bothers me on purely moral grounds if nothing else. | 19:54 |
| cdent | I think pointing to adam's thing is overstating the case. We didn't mock a method. Then we did. | 19:54 |
| cdent | I agree that it shouldn't be necessary | 19:54 |
| cdent | don't look at me, by the way: I did it with normal wsgi-intercept back when it was just test_report_client | 19:55 |
| efried | Oh, I'm not blaming you, or anyone. | 19:55 |
| efried | Obviously it's a finicky, tricky thing, and whoever hacked it up this way (bauzas based on the comments) did a fine thing for the sake of expediency. | 19:56 |
| cdent | I think we should be happy we have thing that does a reasonable job and make cyborg go instead | 19:57 |
| efried | good call | 19:57 |
| cdent | I'm similarly stuck on https://review.openstack.org/#/c/636807/3 | 20:09 |
| *** openstackgerrit has quit IRC | 20:09 | |
| cdent | thinks about it on a piece of paper | 20:24 |
| *** cdent has quit IRC | 20:24 | |
| *** efried has quit IRC | 20:29 | |
| *** efried has joined #openstack-placement | 20:29 | |
| *** tssurya has quit IRC | 20:32 | |
| *** ttsiouts has joined #openstack-placement | 20:36 | |
| *** e0ne has joined #openstack-placement | 20:37 | |
| *** ttsiouts has quit IRC | 21:00 | |
| *** ttsiouts has joined #openstack-placement | 21:01 | |
| *** ttsiouts has quit IRC | 21:05 | |
| *** e0ne has quit IRC | 21:28 | |
| *** openstackgerrit has joined #openstack-placement | 22:04 | |
| openstackgerrit | Merged openstack/placement master: Downgrade os-traits/os-resource-classes sync log level to DEBUG https://review.openstack.org/637892 | 22:04 |
| *** ttsiouts has joined #openstack-placement | 22:07 | |
| *** ttsiouts has quit IRC | 22:12 | |
| *** mriedem has quit IRC | 23:41 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!