Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2020-09-21
17:11:34 artom dansmith, how though? I thought we established we can't have other tests importing eventlet before us
17:11:41 dansmith if you don't discovery and don't run anything else
17:11:48 sean-k-mooney artom: i think this is where we currently monkeyp[atch for test by the way https://github.com/openstack/nova/blob/master/nova/test.py#L24
17:12:08 dansmith artom: I still mean a special invocation like you have, but a unittest instead of a bare python file
17:12:29 sean-k-mooney artom: so when youd do "from nova import test"
17:12:37 sean-k-mooney that shoudl monkey patch
17:12:48 artom Right, but unittest means stestr will read the whole tree... and so will cause eventlet to be imported, even if we don't execute the tests themselves
17:13:02 dansmith no,
17:13:05 dansmith not without discover I think
17:13:50 dansmith for other unittest runners, if you provide an actual class module without discovery, it will try to import that path direct and then execute
17:13:51 artom Oh, you literally mean `--no-discover`
17:13:52 sean-k-mooney dansmith: we would have to put it in a seperate folder
17:13:58 dansmith artom: right, like the profile test does
17:14:13 artom *facepalm* OK yeah
17:14:15 sean-k-mooney we could skip using a regex if we did that and use the --isolate parmater i think
17:14:24 dansmith sean-k-mooney: no, you can just make it skip for the regular tests, if necessary, or let it run in the regular ones if it will falsely claim success anyway :)
17:14:25 artom Worth a try, anyways
17:14:40 dansmith sean-k-mooney: you could just make it skip if sys.modules already has eventlet
17:14:54 sean-k-mooney dansmith: ah
17:14:55 artom This bare Python file is definitely teh uglies
17:15:06 sean-k-mooney i tought you were going to sugges adding it to the test blacklist
17:15:13 sean-k-mooney but having the test auto skip i think is fine
17:15:31 sean-k-mooney that way we dont need to list a load of special tests if we add more like htis
17:15:40 dansmith right
17:15:41 artom Auto-skip is dangerous, no? Means we'd be silently missing potential errors
17:15:52 sean-k-mooney no in the ci
17:15:55 sean-k-mooney locallly sure
17:16:03 dansmith artom: well, you can also do what the profile test does and set an envar to enable :)
17:16:05 sean-k-mooney but if its run by tox
17:16:11 sean-k-mooney it should be fine
17:16:39 artom dansmith, ah, so folks can disable it if needed
17:16:50 artom Whodda thunk precedent was so useful ;)
17:16:51 sean-k-mooney artom: proably enable if needed
17:16:52 dansmith artom: no,
17:16:59 sean-k-mooney given its racy by defualt
17:17:12 dansmith artom: it's disabled by default, enabled explicitly in that second stestr run command in tox
17:17:14 sean-k-mooney you would set the env var wehn you run it explcitly
17:17:25 sean-k-mooney yep what dansmith said
17:17:28 artom dansmith, sorry, right
17:17:49 artom I meant in the sense of... tox will run with it enabled
17:18:03 artom Everything else won't
17:18:06 sean-k-mooney tox -e py<whatever> sure
17:18:19 dansmith seriously, there are a bunch of other things in that file with comments like "this is fragile if order isn't respected" comments... it'd be nice to get tests for those things too since we're clearly just flying by the seat of our pants on these things
17:18:26 sean-k-mooney but in two calls to stester
17:18:47 artom Wait, no, that second tox command that I'll add will enable and run it
17:18:53 artom Nothing else will because it's racy
17:19:01 dansmith urllib3, oslo_context, threading things,
17:19:10 dansmith artom: right, exactly.. you know, like the profiling test :)
17:19:21 sean-k-mooney dansmith: ya we have a cople of example of this
17:19:39 artom dansmith, you keep saying that like I'm some sort of genius that understands it the first time around
17:19:41 artom ;)
17:19:53 sean-k-mooney not sure why oslo_context would be on that list but why not
17:19:57 dansmith artom: it just makes me feel better to highlight the first-mention-to-grok time gap :)
17:20:09 dansmith sean-k-mooney: maybe if we had a test with comments in it, we'd know :/
17:20:33 artom dansmith, I'm pretty sure it's exponential with age (and inverse with sleep )
17:20:40 sean-k-mooney :) well it was more oslo is an openstack thing so hopefully we could fix the oslo issue there instead of nova
17:20:47 dansmith artom: we should plot it on a graph
17:21:14 dansmith sean-k-mooney: not likely if it's something oslo can't do if imported post-monkey-patch
17:21:16 artom ∞ there's your graph
17:21:19 dansmith heh
17:22:41 openstack Launchpad bug 1773102 in OpenStack Compute (nova) queens "Abnormal request id in logs" [Medium,Fix committed] - Assigned to Radoslav Gerganov (rgerganov)
17:22:41 sean-k-mooney https://bugs.launchpad.net/nova/+bug/1773102
17:22:51 sean-k-mooney thats the oslo_context bug
17:23:20 dansmith ah yeah, and surely that's because it needs to grab pointers to the threading primitives for TLS before they get munched right?
17:24:06 sean-k-mooney maybe although there might be a way to resolve that in a differnt way. anyway not important right now
17:24:51 sean-k-mooney artom: so you going to add a second run combining the results with teh first and explictly enable that test in the second run ya
17:24:53 openstack Launchpad bug 1773102 in OpenStack Compute (nova) queens "Abnormal request id in logs" [Medium,Fix committed] - Assigned to Radoslav Gerganov (rgerganov)
17:24:53 dansmith https://bugs.launchpad.net/nova/+bug/1773102/comments/18
17:25:00 dansmith it'd be hard to solve it any other way, but.. yah
17:25:35 artom dansmith, for all this talk of "do it like the profiler", I've tried it, and looks like it's still importing eventlet elsewhere
17:26:20 sean-k-mooney artom: can you push it so we can try it
17:29:04 openstackgerrit Artom Lifshitz proposed openstack/nova master: Unit test for 7c1d964faa https://review.opendev.org/753072
17:29:08 artom sean-k-mooney, ^^
17:31:38 dansmith artom: can you try it with python3 -munittest path.to.test ?
17:32:35 dansmith artom: wait, you're still importing nova.test in your module
17:32:36 dansmith that means you do it before setUp() runs :)
17:33:00 sean-k-mooney yep that would do it
17:33:04 dansmith so that "importing eventlet elsewhere" is ... in your test :)
17:33:52 artom Right, but nova.test imports eventlet *after setting the greendns env var*
17:34:01 artom So that should be fine
17:34:21 artom The problem is if some other unit tests straight up `import eventlet` without setting EVENTLET_NO_GREENDNS
17:34:30 dansmith I thought something else does that is downstream of that, and that was the whole point?
17:34:46 artom Something else does what?
17:34:58 dansmith import eventlet
17:35:24 sean-k-mooney that what we get when we run it directly http://paste.openstack.org/show/798154/
17:36:23 artom Weird
17:38:11 sean-k-mooney basically we need to run this test using unitt test and not the base testcase and only import monkey patch inside teh test funtion i think
17:38:19 sean-k-mooney im going to try that locally
17:38:41 dansmith right
17:38:58 dansmith although I moved the import into the test function and I still fail the test
17:39:27 dansmith so, here's the other thing
17:39:41 dansmith stestr forks and runs the worker children outside the main process
17:39:57 dansmith which may not share the environment where the flag is being set
17:40:09 artom Sounds "right" - in the sense that, if we're the *only* ones importing nova.test, and that imports eventlet "properly" (setting EVENTLET_NO_GREENDNS), shouldn't matter when we do the importing
17:41:04 dansmith still happens with bare unittest run
17:41:27 sean-k-mooney http://paste.openstack.org/show/798157/
17:42:07 sean-k-mooney so it match on the function call
17:42:24 artom *facepalm*
17:42:49 sean-k-mooney test_greendns_is_disabled
17:42:50 dansmith ah hah

Earlier   Later