| openstackgerrit | Merged openstack-infra/shade: Update test of object metadata to mock requests https://review.openstack.org/411911 | 00:07 |
|---|---|---|
| openstackgerrit | Merged openstack-infra/shade: Change get_object_metadata to use REST https://review.openstack.org/411912 | 00:07 |
| openstackgerrit | Monty Taylor proposed openstack-infra/shade: Replace mocks of swiftclient with request_mock https://review.openstack.org/412108 | 00:59 |
| *** gouthamr has joined #openstack-shade | 02:56 | |
| *** gouthamr has quit IRC | 03:11 | |
| *** noama has quit IRC | 03:12 | |
| *** noama has joined #openstack-shade | 03:12 | |
| *** abregman has joined #openstack-shade | 06:12 | |
| *** abregman has quit IRC | 06:31 | |
| *** abregman has joined #openstack-shade | 06:36 | |
| *** yfried has joined #openstack-shade | 08:31 | |
| *** noama has quit IRC | 08:48 | |
| *** cdent has joined #openstack-shade | 11:11 | |
| *** yfried has quit IRC | 11:25 | |
| *** yfried has joined #openstack-shade | 11:25 | |
| *** yfried has quit IRC | 11:34 | |
| *** yfried has joined #openstack-shade | 11:34 | |
| *** yolanda has quit IRC | 11:59 | |
| *** yolanda has joined #openstack-shade | 12:00 | |
| *** gouthamr has joined #openstack-shade | 13:26 | |
| *** cdent has quit IRC | 14:16 | |
| mordred | Shrews, morgan: so - I'm about 4 test cases away from having converted all of the swift test cases to requests_mock. It's actually been both fun and informative - and I've got notes in the patch about things we should change in the next patch (which should convert from swiftclient to rest) | 14:33 |
| *** yfried has quit IRC | 14:44 | |
| *** yfried has joined #openstack-shade | 14:46 | |
| mordred | HOLY CRAP I'M DONE!!!! | 14:48 |
| openstackgerrit | Monty Taylor proposed openstack-infra/shade: Replace mocks of swiftclient with request_mock https://review.openstack.org/412108 | 14:48 |
| mordred | Shrews: that ^^ converts the testing of all of the swiftclient calls (although it does not do SwiftService) | 14:49 |
| * Shrews solving the ENOCOFFEE error atm | 14:50 | |
| mordred | Shrews: it's the most important problem to solve | 14:52 |
| *** cdent has joined #openstack-shade | 14:55 | |
| *** jamielennox is now known as jamielennox|away | 14:56 | |
| *** cdent has quit IRC | 16:26 | |
| *** abregman has quit IRC | 16:35 | |
| morgan | mordred: woot | 16:49 |
| morgan | mordred: niiice | 16:49 |
| morgan | mordred: i am looking for my laptop in the pile of boxes | 16:49 |
| morgan | mordred: but i am now in seattle and have good internet (tm) | 16:50 |
| * morgan is typing on underpowered netbook that can't run unit tests | 16:50 | |
| mordred | morgan: woot | 16:52 |
| dtroyer | mordred: I just noticed your "swiftclient to REST" comment above, are you planning to re-implement those directly in shade? | 16:54 |
| *** cdent has joined #openstack-shade | 16:56 | |
| openstackgerrit | Monty Taylor proposed openstack-infra/shade: Stop calling HEAD before DELETE for objects https://review.openstack.org/412528 | 16:56 |
| openstackgerrit | Monty Taylor proposed openstack-infra/shade: Stop using full_listing in prep for REST calls https://review.openstack.org/412529 | 16:56 |
| openstackgerrit | Monty Taylor proposed openstack-infra/shade: Switch swift calls to REST https://review.openstack.org/412530 | 16:56 |
| mordred | dtroyer: yah - the dependencies on the client libs turns out to be an issue for folks - since shade wants newer client libs, it makes installing it near an older install of openstack problematic | 16:57 |
| dtroyer | you know OSC has had a subset of swiftclient built-in for 4 years now… :) | 16:58 |
| mordred | dtroyer: glance is done - have everything for swift now except for create_object because we're using SwiftService which is going to take some more work | 16:58 |
| mordred | dtroyer: :) | 16:58 |
| morgan | dtroyer: and keystone is on the short-list | 16:58 |
| mordred | dtroyer: I'm finding that direct rest is WAY easier to deal with than the client libs | 16:58 |
| dtroyer | so, politically incorrect question, want to make a stand-aline low-level SDK? | 16:58 |
| dtroyer | what is in OSC is what I proposed for that for the Python SDK and was ignored | 16:58 |
| morgan | dtroyer: i'd be open to such things, but first priority is getting shade working this way | 16:58 |
| dtroyer | I dont' want an ORM, I just want that too :) | 16:58 |
| mordred | dtroyer: I would ... except honestly that sdk is just keystone auth - the only wrapper stuff I've got in shade right now is munchifcation | 16:59 |
| morgan | dtroyer: but for the most part... what mordred just said | 16:59 |
| dtroyer | ok, just curious. | 16:59 |
| dtroyer | we keep duplicating efforts... | 16:59 |
| morgan | dtroyer: getting shade 100% rest opens doors for other things (inc. replicating the work for low level sdk things) | 16:59 |
| mordred | dtroyer: (my original thought before I started doign this is pretty much exactly what you just said - but it's shockingly easy to deal with so far... I'm surprised we ever wrote libraries) | 17:00 |
| mordred | dtroyer: and yes on the duplicating efforts front | 17:00 |
| morgan | i would rather see shade do the work directly first, since that is the priority work. | 17:00 |
| dtroyer | you guys run along ahead since you have multiple folk working on it, I'm heading down a different client route already | 17:00 |
| dtroyer | give you 3 gueses as to what | 17:00 |
| morgan | :) | 17:00 |
| mordred | dtroyer: :) | 17:00 |
| morgan | hehe | 17:00 |
| mordred | dtroyer: maybe I get you an oaktree server it can talk to too :) | 17:01 |
| * dtroyer would be in dreamland | 17:01 | |
| morgan | mordred: you know what the best part of seattle is? | 17:01 |
| morgan | mordred: same day prime shipping. | 17:01 |
| mordred | morgan: you didn't have that in portland? | 17:02 |
| morgan | mordred: "oh i need more ethernet cables" | 17:02 |
| mordred | morgan: I have that here in dallas - it changes everything | 17:02 |
| morgan | mordred: nope. never. in fact didn't have that in pasadena either | 17:02 |
| mordred | I almost never buy things anymore if I can't get them same-day primed | 17:02 |
| morgan | mordred: but omg... "i want this thing now" is dangerous | 17:02 |
| mordred | yup | 17:02 |
| dtroyer | "here in Dallas" mordred did you move (again)? | 17:03 |
| morgan | also prime fresh... will be amazing for this week while still unpacking | 17:03 |
| morgan | dtroyer: mordred has lived in dallas for... a bit now | 17:03 |
| morgan | dtroyer: i just moved to seattle from PDX | 17:03 |
| * dtroyer can't keep up, been in same house for 9 years now | 17:04 | |
| morgan | as soon as i have a down payment... planning on buying a place either here or in nyc. | 17:04 |
| mordred | dtroyer: yah - like a year ago ... we bought a house here so I'm hoping this is the last move for a while | 17:04 |
| morgan | depending on some things (depending on job things, not mine) | 17:05 |
| morgan | mordred: i think i'll have that neutron fixture thing done today | 17:05 |
| morgan | and on to keystone rest things | 17:05 |
| morgan | mordred: did you get the context manager example up? | 17:06 |
| morgan | for ksa? | 17:06 |
| mordred | morgan: nope! because I suck | 17:06 |
| morgan | mordred: totally unrelated, white rabbit project in 4k is pretty darn awesome | 17:07 |
| mordred | morgan: s/white rabbit project/everything/ | 17:07 |
| morgan | i might eat my entire 1TB (sigh comcast) bandwidth on 4k streaming | 17:07 |
| morgan | yeah that too | 17:07 |
| morgan | also $2 for 4k on netflix is not a bad deal | 17:08 |
| mordred | right? | 17:15 |
| openstackgerrit | Monty Taylor proposed openstack-infra/shade: Switch swift calls to REST https://review.openstack.org/412530 | 17:26 |
| openstackgerrit | Monty Taylor proposed openstack-infra/shade: Make delete_object return True and False https://review.openstack.org/412543 | 17:26 |
| openstackgerrit | Monty Taylor proposed openstack-infra/shade: Update swift exception tests to use 416 https://review.openstack.org/412544 | 17:26 |
| thingee | hey all. back from paternity leave :) | 17:33 |
| morgan | thingee: welcome back! | 17:41 |
| openstackgerrit | Monty Taylor proposed openstack-infra/shade: Make assert_calls a bit more readable https://review.openstack.org/412549 | 17:43 |
| mordred | yay it's a thingee !!! | 17:43 |
| openstackgerrit | Merged openstack-infra/shade: Add release notes and an error message for release https://review.openstack.org/411829 | 17:52 |
| openstackgerrit | Morgan Fainberg proposed openstack-infra/shade: Add in a dynamic project scope test fixture https://review.openstack.org/410453 | 18:17 |
| morgan | mordred: ^ just need the scope override bits. now | 18:18 |
| morgan | mordred: (and the forced re-auth, clearly) | 18:18 |
| mordred | woot | 18:21 |
| mordred | morgan: I gotta run to the store - will poke when I get back | 18:22 |
| morgan | mordred: sounds good. | 18:22 |
| openstackgerrit | Morgan Fainberg proposed openstack-infra/shade: Add in a dynamic project scope test fixture https://review.openstack.org/410453 | 18:24 |
| *** yfried has quit IRC | 18:25 | |
| thingee | oh yay repo merge | 19:02 |
| thingee | so the comment from the bot says it's automatically moving stuff over, but issues like this one I can't find in ansible/ansible https://github.com/ansible/ansible-modules-core/issues/4108#issuecomment-262195852 | 19:03 |
| *** openstackstatus has quit IRC | 19:13 | |
| *** openstack has joined #openstack-shade | 19:13 | |
| morgan | mordred: when you're back need some help figuring out where to hook in the scope change. I'm thinking that we should supply a new auth-plugin with the changed values... but it only works with password/v3password that wya | 19:15 |
| mordred | morgan: well, at the moment we have no functional tests of anything other than password plugins, because as of yet I haven't figured out how to do devstack that way :) | 19:17 |
| morgan | are we also ok with this fixture being v3 only? | 19:18 |
| morgan | because if so... | 19:18 |
| morgan | that makes this way easier | 19:18 |
| morgan | i'll just supply a new session for shade when we enter the context for the fixture | 19:19 |
| morgan | and restore the original one when we're done | 19:19 |
| mordred | yah - I think that'll be fine | 19:19 |
| morgan | cool. then almost done, maybe another hour | 19:19 |
| morgan | then we need to test it / use it | 19:19 |
| mordred | morgan: I mean - we're going to have to do a little bit of work to make sure we're not trying to run tests that do this when we're running in keystonev2 backwards compat cloud | 19:19 |
| mordred | morgan: but maybe the fixture can detect that env var and throw a skipTest if it finds it | 19:20 |
| *** yfried has joined #openstack-shade | 19:20 | |
| mordred | morgan: SHADE_USE_KEYSTONE_V2 is the env var that will be set if we're testing keystone v2, fwiw | 19:20 |
| morgan | yah. once we're sure it works we can be a bit pickier | 19:20 |
| morgan | actually... | 19:20 |
| morgan | i can just look at the plugin type and supply the information | 19:21 |
| morgan | i already do so.. so it can do v2 and v3 | 19:21 |
| mordred | ++ | 19:30 |
| mordred | morgan: where does identity_version come from? | 19:32 |
| mordred | oh - that's an instance of a test, not an instance of a cloud | 19:32 |
| mordred | nm | 19:32 |
| morgan | yep | 19:35 |
| openstackgerrit | Morgan Fainberg proposed openstack-infra/shade: Add in a dynamic project scope test fixture https://review.openstack.org/410453 | 19:45 |
| morgan | mordred: ok there we go | 19:45 |
| morgan | that should do it... I hope | 19:45 |
| openstackgerrit | Morgan Fainberg proposed openstack-infra/shade: Add in a dynamic project scope test fixture https://review.openstack.org/410453 | 19:54 |
| morgan | typo fix | 19:54 |
| *** cdent has quit IRC | 20:36 | |
| mordred | morgan: wow, there's a bunch of stuff in there! | 21:01 |
| *** jamielennox|away is now known as jamielennox | 21:04 | |
| mordred | morgan: so ... | 21:10 |
| mordred | morgan: self.use_fixture(scoped_auth.DynamicProjectScopedAuth(self)) | 21:10 |
| mordred | that's about what I should do to use it, right? | 21:10 |
| openstackgerrit | Monty Taylor proposed openstack-infra/shade: Add functional test to show floating ip reuse https://review.openstack.org/409842 | 21:12 |
| mordred | morgan: ^^ here goes nothing | 21:12 |
| morgan | mordred: that is the idea. | 21:13 |
| morgan | let's see if this works. :) | 21:14 |
| mordred | I was just typing the same thig | 21:14 |
| *** gouthamr has quit IRC | 21:17 | |
| *** abregman has joined #openstack-shade | 21:23 | |
| mordred | jamielennox: morning! so I did a big requests_mock shade patch over the weekend: https://review.openstack.org/#/c/412108/ and while doing it ran across something (in that commit message) that could either be a swiftclient bug or a requests_mock bug | 21:51 |
| mordred | jamielennox: my hunch (and my one of my next tasks is to try to prove this one way or the other) is that it's a swiftclient bug - but I thought I'd show you in case you were like "oh oops" | 21:51 |
| jamielennox | mordred: that's a big change - what am i looking at/ | 22:02 |
| mordred | oh - sorry ... tl;dr is more the bit in shade/openstackcloud.py where I had to prefix the two strings with b' | 22:03 |
| mordred | jamielennox: if I didn't, I would get b'' back from the fake rest call, and then try to compare the two and they wouldn't match. I _think_ swiftclient is doing the conversion | 22:04 |
| jamielennox | mordred: so i know i've looked at converting swiftclient to requests-mock in the past and i just found a disaster of string processing and test mocking | 22:05 |
| *** abregman has quit IRC | 22:05 | |
| jamielennox | so their tests are completely unreliable in testing what they do | 22:05 |
| mordred | jamielennox: awesome. well, at least it isn't just me | 22:06 |
| jamielennox | on the other hand, i'm not sure off the top of my head what the correct thing to do with str vs bytes in python3 for http headers | 22:06 |
| jamielennox | i think requests makes them string | 22:06 |
| jamielennox | but i would need to double check | 22:06 |
| mordred | jamielennox: well -the swift api docs make some mention that headers are are supposed to be utf8 or something - lemme find the text | 22:07 |
| jamielennox | from a requests-mock perspective i think whatever you put into the response dictionary is what you would get out | 22:07 |
| jamielennox | i don't think it messes with type encoding at all - unless the requests header dict is doing something | 22:08 |
| mordred | jamielennox: that, to me, seems to be what I'd expect as a user | 22:10 |
| jamielennox | i can probably step through that and see what's happening | 22:10 |
| jamielennox | shade is obviously doing this against a real server so i'm guessing it's right | 22:11 |
| openstackgerrit | Monty Taylor proposed openstack-infra/shade: Add functional test to show floating ip reuse https://review.openstack.org/409842 | 22:13 |
| mordred | yah - but we don't exercise the fringes of text/bytes/utf-8 in http-headers by any means - so we could be doing something "wrong" that's close enough | 22:14 |
| mordred | morgan: ^^ typo fix :) | 22:14 |
| *** yfried has quit IRC | 22:15 | |
| morgan | mordred: ahh | 22:19 |
| mordred | jamielennox: I took the simple route and went and read the swiftclient source code | 22:35 |
| jamielennox | mordred: you thought that was simple until you opened it | 22:36 |
| mordred | jamielennox: heh | 22:36 |
| mordred | resp_header_dict calls parse_header_string | 22:36 |
| mordred | jamielennox: which does a .decode('utf8') | 22:36 |
| mordred | which explains why it shows up as bytes in python3 | 22:36 |
| jamielennox | which means requests-mock caught a bug? | 22:38 |
| mordred | well - they also encode to utf8 when you write the headers | 22:38 |
| mordred | but yeah - I'm not sure I agree it's the right thing to do - but I'm also not sure they could change it on their users at this point | 22:39 |
| jamielennox | it's weird for shade though, because that check would never have worked ? | 22:48 |
| jamielennox | actually i guess it just would have just failed on python 3 - which honestly no one would probably notice | 22:48 |
| mordred | yah. I think that's the real thing - nobody is actually using python3 in anger yet | 22:49 |
| mordred | morgan: "AttributeError: 'TestFloatingIP' object has no attribute 'identity_version'" | 22:58 |
| mordred | http://logs.openstack.org/42/409842/9/check/gate-shade-dsvm-functional-neutron/e8c6119/console.html | 22:58 |
| mordred | morgan: you may want to pull self.demo_cloud.cloud_config.get_api_version('identity') | 23:00 |
| morgan | mordred: ahh | 23:01 |
| morgan | mordred: will take a look at that in a sec | 23:01 |
| morgan | mordred: fixing that bug now | 23:32 |
| *** gouthamr has joined #openstack-shade | 23:58 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!