Wednesday, 2014-09-24

*** mfer has joined #openstack-sdks00:54
*** mfer has quit IRC00:59
*** briancurtin has joined #openstack-sdks01:46
*** briancurtin has joined #openstack-sdks01:46
*** rmcall has quit IRC02:43
*** rmcall has joined #openstack-sdks03:21
openstackgerritwanghong proposed a change to openstack/python-openstackclient: utils.find_resource does not catch right exception
*** stevemar has joined #openstack-sdks03:46
openstackgerritRodrigo Duarte proposed a change to openstack/python-openstackclient: Enables retrieval of project's parents and subtree
*** rmcall has quit IRC04:25
*** rmcall has joined #openstack-sdks04:27
*** stevemar has quit IRC04:54
*** stevemar has joined #openstack-sdks04:54
*** rmcall has quit IRC04:58
*** stevemar has quit IRC05:31
*** k4n0 has joined #openstack-sdks06:09
*** ayoung has quit IRC06:19
*** ayoung has joined #openstack-sdks06:40
*** briancurtin has quit IRC06:53
*** jamiehannaford has joined #openstack-sdks07:41
*** openstack has joined #openstack-sdks09:24
*** jamiehan_ has joined #openstack-sdks09:38
*** jamiehannaford has quit IRC09:41
*** jamiehan_ has quit IRC09:43
*** jamiehannaford has joined #openstack-sdks09:46
openstackgerritOpenStack Proposal Bot proposed a change to openstack/python-openstackclient: Updated from global requirements
*** HenryG_afk is now known as HenryG12:09
*** ayoung has quit IRC12:57
*** mfer has joined #openstack-sdks12:57
*** bknudson has joined #openstack-sdks13:30
*** samuelmz has joined #openstack-sdks13:49
samuelmzHi, it looks like python-openstackclient cannot use identity api v3 anymore ...13:53
samuelmzSince the patch
samuelmzWe are getting this error:13:54
samuelmzBadRequest: Expecting to find domain in project - the server could not comply with the request since it is either malformed or otherwise incorrect. The client is assumed to be in error. (HTTP 400)13:54
samuelmzIf we revert it, we are able to use identity v3 again ..13:54
*** briancurtin has joined #openstack-sdks13:59
samuelmzdolphm, bknudson, lbragstad ping14:01
*** sigmavirus24_awa is now known as sigmavirus2414:02
samuelmzWhat are your thoughts on this ^? What should we do? Do we revert the change or should we open a new bug to fix that?14:03
lbragstadsamuelmz: o/14:04
lbragstadsamuelmz: what call are you making?14:04
samuelmzI got this error with any call ..14:06
samuelmze.g. just `openstack` give the error14:06
samuelmzlbragstad, I just ran devstack and changed my openrc to get a token set to use identity v314:07
samuelmzlbragstad, and then: `. openrc admin admin`14:08
samuelmzlbragstad, `openstack`14:08
lbragstadsamuelmz: stevemar would probably be a good resource to ask, he's done a lot of work with keystone v3 and the openstack cli14:11
samuelmzlbragstad, ok .. I'm gonna ask him once he is up14:13
samuelmzlbragstad, thanks14:13
lbragstadsamuelmz: no problem, I don't see him in -keystone but I'll keep an eye out for him too14:13
samuelmzlbragstad, I will also ask dtroyer once he's up .. he commited that change .. he may know what impact that had and then find what's happening :-)14:15
lbragstadsamuelmz: good idea14:15
terrylhowethe 400 is because the user and project domain weren’t specified, right lbragstad ?14:24
terrylhoweI think you should be able to auth if you just specifiy the user and project domain, I don’t have devstack up right now to try it14:26
samuelmzterrylhowe, gonna try this ..14:32
*** stevemar has joined #openstack-sdks14:40
samuelmzterrylhowe, I tried14:40
samuelmzterrylhowe, openstack --os-auth-url localhost:5000/v3 --os-username admin --os-user-domain-name Default --os-password admin --os-project-name admin --os-project-domain-name Default --os-region-name RegionOne14:40
samuelmzterrylhowe, and got 'ERROR: Exception raised:'14:41
samuelmzterrylhowe, without further information ..14:41
terrylhowe—debug ?14:41
samuelmzterrylhowe, am I missing any info when getting the token?14:41
samuelmzoh wait14:42
samuelmzlooks like didn't set identity api version14:42
terrylhowev2 auth14:42
samuelmzjust appended `--os-identity-api-version 3` and got the same error14:44
terrylhowewell, it is using v3 auth at least14:45
terrylhowesamuelmz: how about making auth_url a real url?14:51
terrylhoweI can’t recreate what your seeing14:51
dtroyer[catching up]14:51
samuelmzterrylhowe, I had forgotten 'http://' on --os-auth-url14:52
terrylhoweI have a trailing slash on mine too14:52
samuelmzterrylhowe, openstack --os-auth-url http://localhost:5000/v3 --os-username admin --os-user-domain-name Default --os-password admin --os-project-name admin --os-project-domain-name Default --os-region-name RegionOne --os-identity-api-version 314:52
samuelmzterrylhowe, ^it works14:53
samuelmzterrylhowe, hi dtroyer14:53
samuelmzhi dtroyer*14:53
terrylhoweahh, cool :)14:53
dtroyerso is there a change in behaviour?  it looks like originally a command line that used to work doesn't any more14:54
terrylhowehe just got it working dtroyer14:55
terrylhoweauth-url wasn’t a full url and possible missing domains for v314:55
samuelmzdtroyer, in fact it works .. but now we need to explicitely set --os-user-domain-name and --os-project-domain-name when using devstack14:56
*** jamiehan_ has joined #openstack-sdks14:57
dtroyerok, that's what I was wondering.14:57
samuelmzterrylhowe, dtroyer, yes we get it working .. but 1) before we didn't need to set --os-user-domain-name and --os-project-domain-name when using identity v3 ... and now we do14:57
dtroyerthos should both be 'Desault' as you set…need to figure out if that is a devstack thing or osc thing14:57
terrylhowethat is a devstack thing14:58
dtroyerok, thats an osc change.  probably due to calling the ksc auth plugins directly, they seem to be pickier than the old auth code14:58
dtroyerI'd bet ksc was setting those for us before14:59
*** jamiehannaford has quit IRC14:59
terrylhoweor keystone changed14:59
dtroyerit may have, but we drastically changed how we call it14:59
terrylhoweI’d like to get with Jamie when he gets back and get an understanding of the different ways you can auth with the plugins15:01
dtroyerand it'll change again soon with Matthieu's support for arbitrary plugins15:02
dtroyerterrylhowe: one of the problems with the plugins as they are in ksc right now is that there is stull back-compat stuff in there that I really want to avoid.  and it isn't always obvious what that is…15:03
dtroyersamuelmz: if you have a minute could you try and see if it has the same issue?15:04
samuelmzdtroyer, yes sure15:04
samuelmzdtroyer, yes I have the same issue ... but I can't say it introduced the same thing (by itself) as yours because yours is also in my git log ...15:07
samuelmzdtroyer, makes sense?15:07
dtroyeryes, they are dependent.  I was curious if the change in how the plugins are called made a difference15:08
samuelmzdtroyer, actually I don't see any problem in having to provide  --os-user-domain-name and --os-project-domain-name from now15:08
samuelmzdtroyer, I don't see how we was doing this before ...15:09
dtroyerI think ksc's old-client compatability was doing it for us15:10
samuelmzdtroyer, we can have a user and a project with same name in different domains .. so for me it's obvious that we need to provide those parameters ..15:10
dtroyersure, but sensible defaults, say matching what keystone does, would be nice15:10
samuelmzdtroyer, I got it .. v2 = no domains -> 'Default'15:11
samuelmzdtroyer, so if we use v2 and do not provide those parameters ... it do for us .. right?15:12
samuelmzdtroyer, but if we use v3, it doesn't have to set domain to 'Default'15:13
dtroyersamuelmz: yes.  my understanding is that then directly using the v3 endpoint it should still work, if we supply the default names now15:13
samuelmzdtroyer, we must set it explicitely15:13
openstackgerritDean Troyer proposed a change to openstack/python-openstackclient: Set default domains
dtroyersamuelmz: this ^^^ is a quick patch to set those.  both of my devstacks are deep into odd configus ATM so I haven't tested it yet15:14
samuelmzdtroyer, gonna try .. almost sure it will work15:14
*** jamiehan_ has quit IRC15:15
samuelmzdtroyer, commented on your patch15:20
samuelmzdtroyer, if project_domain_name may come as an empty string ..15:20
dtroyersamuelmz: first reaction is that may be intentional…but thinking about it I don't think that is ever valid, right?15:22
dtroyerBTW, I don't plan to merge that, whatever happens will be in the context of which handles the clientmanager args quite differently15:23
samuelmzdtroyer, ok ..15:24
terrylhowewell the env function I think does return ‘’ by default doesn’t it dtroyer15:24
dtroyerah, right, it does15:25
* samuelmz is hungry and he's gonna have lunch .. back in few minutes15:27
*** briancurtin has quit IRC15:28
*** briancurtin has joined #openstack-sdks15:29
*** briancurtin2 has joined #openstack-sdks15:30
*** ayoung has joined #openstack-sdks15:30
*** briancurtin has quit IRC15:33
lbragstadsamuelmz: stevemar is around now :)15:37
lbragstadsamuelmz: not sure if your questions were completely addressed or not15:38
stevemarlbragstad, always assume questions have been addressed, it allows you to go back to guilt-free work!15:39
dtroyerand if anyone knows about guilt-free work, it's stevemar15:42
stevemardtroyer, it's the best kind15:44
stevemardtroyer, i'm getting pressure about osc again, folks keep asking if it's ready for prime time15:44
dtroyerDo you think it is?  There are a lot of things left to do and clean up yet, but what is there is pretty solid IMnsHO15:46
stevemari'm mostly of the same opinion15:46
stevemardtroyer, we cover the first layer of the 'big tent' pretty well15:47
stevemarwith the exception of designate15:47
*** briancurtin2 is now known as briancurtin15:48
*** briancurtin has joined #openstack-sdks15:48
dtroyerand neutron.  Terry's done a bunch of work but there's sooo much more to that bugger15:48
stevemardtroyer, yeah... neutron *shudders*15:49
dtroyerstevemar: when you get a minute I'd like to know what your thoughts on and are.15:57
*** k4n0 has quit IRC16:03
*** stevemar has quit IRC16:20
*** openstackgerrit has quit IRC16:35
samuelmzdtroyer, so you have a lot of interesting work to be done on osc ? :-)16:49
dtroyersamuelmz: tons of work…some of it interesting16:50
dtroyerwe have a significant shortage in test coverage16:50
dtroyerand a lot of missing commands, such as the entire volume v2 set16:51
samuelmzdtroyer, hm .. I thing the test coverage thing is interesting ..16:51
samuelmzdtroyer, you talk about a specific service? or we have lacks of coverage in all services ..16:52
dtroyersamuelmz: take a look at, it's a new approach that includes the api lib in the tests16:52
dtroyerI think we want something like that for the functional tests stevemar is working on16:53
*** jamielenz has joined #openstack-sdks16:53
*** jamielenz is now known as jamielennox16:53
dtroyerfor object-store commands  we're also the low-level api16:54
*** dtroyer has quit IRC16:54
*** dtroyer has joined #openstack-sdks16:56
*** sigmavirus24 is now known as sigmavirus24_awa16:57
*** stevemar has joined #openstack-sdks17:02
*** f13o has quit IRC17:03
samuelmzdtroyer, something to be done with keystone part of osc? (related to coverage)17:08
*** mfer has quit IRC17:08
samuelmzdtroyer, I'm more familiar with keystone ... it'd be easier to start ..17:08
dtroyersamuelmz: stevemar is the guy to ask there, he's done the bulk of that work17:09
stevemarsamuelmz, try out the v3 api's :)17:09
stevemarsamuelmz, i think we are lacking v3 identity tests too17:09
samuelmzstevemar, ok .. I'm gonna take a look at that .. :)17:10
stevemardtroyer, it's getting harder and harder to suggest work items for folks :(17:10
*** mfer has joined #openstack-sdks17:11
samuelmzstevemar, I've a question about mocks, fakes ... : -)17:12
stevemarsamuelmz, sure dtroyer ^17:12
samuelmztake a look at test test_project_show_parents on
dtroyera) fill-in test coverage; b) audit for new commands/options (hasn't been done since May 2013); c) update wiki; d) network, without the neutronclient baggage if possible; e) handle duplicates between network and compute cleanly; f) give me a minute…17:13
samuelmzI'm mocking projects_mock.get.return_value with my fake var called project, right?17:13
samuelmzdtroyer, haha, nice guidelines :)17:14
dtroyerthat doesn't even mention docs or the low-level api refactor17:14
samuelmzstevemar, dtroyer: in that case, at line 842 for instance, what is the difference between doing 1) identity_fakes.PROJECT_WITH_GRANDPARENT['description'] or 2) project['description'] ?17:15
dtroyersamuelmz: none really, unless project were to get munged during the command run, which happens on some commands17:17
dtroyerspecfically ones that rename tenant_id to project_id for example17:17
stevemarsamuelmz, yeah identity_fakes.PROJECT_WITH_GRANDPARENT should remain unchanged if you are using that, like dtroyer says, depends on if you are changing project17:21
samuelmzyeah both identity_fakes.PROJECT_WITH_GRANDPARENT and var 'project' remain unchanged ..17:22
samuelmzthat's what keystoneclient return to osc .. and then osc does some work ..17:22
samuelmzthe result of that work is what I'm asserting at line 842, right?17:23
samuelmzin fact I'd like to know what approach do you guys prefer.. comparing with the object (ifself) I mocked kc with  (var project)17:23
samuelmzor with thing I've used to set the obj 'project': identity_fakes.PROJECT_WITH_GRANDPARENT ..17:24
*** openstackgerrit has joined #openstack-sdks17:24
samuelmzmy point is that we should forgot identity_fakes.PROJECT_WITH_GRANDPARENT once we did a deepcopy of it17:25
dtroyerI don't have a hard preference, I'm going to say I like to have what sets return_value and what is tested for look the same for test reading purposes, then the differences where the take_action() makes a change is obvious.  But I probalby haven't always done that17:25
*** ayoung has quit IRC17:26
samuelmzdtroyer, ok thanks :)17:29
*** ayoung has joined #openstack-sdks17:41
*** openstackgerrit has quit IRC17:51
*** openstackgerrit has joined #openstack-sdks17:52
*** briancurtin has quit IRC17:57
*** sigmavirus24_awa is now known as sigmavirus2417:57
*** briancurtin has joined #openstack-sdks17:58
*** mfer has quit IRC17:59
*** mfer has joined #openstack-sdks18:02
samuelmzdtroyer, earlier I tested osc by passing user_domain_name and project_domain_name as parameters, right?18:02
samuelmzdtroyer, I was doing `openstack --os-auth-url http://localhost:5000/v3 --os-username admin --os-user-domain-name Default --os-password admin --os-project-name admin --os-project-domain-name Default --os-region-name RegionOne --os-identity-api-version 3`18:02
samuelmzdtroyer, but if I try to run any (like `openstack role list ...` command instead of simply calling `openstack`18:03
samuelmzdtroyer, it still doesn't work18:03
samuelmzdtroyer, here I have a log
samuelmzterrylhowe, ^18:05
samuelmztake a look at line 31, it looks like it persists on using v2 identity api18:06
terrylhoweit is past authentication18:09
terrylhoweI think your creds don’t give you access to roles18:09
terrylhowesamuelmz: is that the case?18:09
dtroyersamuelmz: what do your identity service catalog endpoints look like?18:11
terrylhowegood point dtroyer is is doing a geturls on the service catalog18:14
dtroyerline 26 is the obviols discrepency…it's got an IP that isn't and /v2 in the url18:16
stevemardtroyer, function tests merged :O18:17
dtroyeralso, is v3 present in the version return on '/':   curl http://localhost:5000/18:17
stevemarerr the jenkins job for it anyway18:17
dtroyerstevemar: nice!18:17
stevemarnow to play around! weee18:17
*** rmcall has joined #openstack-sdks18:28
stevemardtroyer, where are you actually passing the session object in for the api/object_store api?18:35
dtroyerthe new stuff?  it's in object/ make_client() when the APIv1 instance is created, it gets instance.session18:37
dtroyerwhich is the shared authenticated session18:38
stevemardtroyer, ahh there it is18:38
stevemarthere was a lot of going down the rabbit hole in those 2 patches18:38
stevemardtroyer, also, is BaseAPIVersion ever used?18:40
dtroyernot in those patches.  the identity api set will be the first in but I'm not sure that is in the test branch yet.18:42
dtroyerit could wait I suppose18:42
dtroyerI'm really more curious about thoughts on the overall approach.18:43
dtroyerI don't expect those to merge as-is18:43
stevemardtroyer, the approach is good18:45
stevemarseems solid18:45
stevemarwhat else do you expect to use the api class? neutron?18:45
dtroyerok, good.  I've worked through a bunch of list commands, and the set of flavor commands, and multiple identity version commands trying to hit the corner cases.18:46
dtroyerhonestly, everything.  this is what I want for the SDK low-level API, but I think I'm alone in that18:46
dtroyerso for OSC, object and network18:46
dtroyerand then we'll see18:46
dtroyermaybe image so we can dump the pyopenssl dependency18:46
dtroyerthat one is a #@$%$^$%@!@$@18:47
stevemardtroyer, i'm happy to see object and network use this, i'd want to leverage the client classes as much as possible18:47
stevemarotherwise we are almost doing the same amount of work18:47
stevemarand then we will *never* catch up to parity18:47
dtroyerstevemar: understood.   I've stopped defining parity as what the project libs do. it's what the project APIs do that is important18:48
dtroyerand for your own reassurrance, identity would be the last to change as it's the base for the session/auth stuffs18:49
stevemardtroyer, meh, that part doesn't bug me :P18:49
stevemardtroyer, i just want adoption and more users18:50
stevemarwhich would mean more people helping out18:50
dtroyeryup.  and hopefully more sanity18:51
dtroyerfwiw, my wife is working on her masters now so I have a lot of evening time available…once I finish the tile in the kitchen18:51
stevemardtroyer, finish up with those tiles then! floor or backsplash?18:52
dtroyerbacksplash.  combination of glass and rough-cut granite.18:52
dtroyerI'll never do the rough-cut again.  getting the grout out of it is a pita18:52
samuelmzdtroyer, sorry for the late answer ..18:53
stevemardtroyer, i was supposed to do our backsplash back in april/may... yeah, it never happened18:53
dtroyersamuelmz: only be sorry if you know how to clean rough-cut granite tile ;)18:53
dtroyerstevemar: this was a august weekend project…18:53
samuelmzdtroyer, I'm testing a new functionality on osc (keystone inherited roles and hierarchical projects) ... I reverted your change for a while18:53
samuelmzdtroyer, :-)18:54
samuelmzdtroyer, once I get that environment back I'll tell you what identity service catalog endpoints looks like :-)18:55
dtroyersamuelmz: np, thanks18:58
*** rmcall has quit IRC19:10
*** rmcall has joined #openstack-sdks19:12
*** rmcall has quit IRC19:12
*** rmcall has joined #openstack-sdks19:38
*** openstackgerrit has quit IRC19:46
*** openstackgerrit has joined #openstack-sdks19:47
*** mfer has quit IRC19:56
stevemarterrylhowe, ping20:05
*** mfer has joined #openstack-sdks20:12
*** mfer has quit IRC20:19
*** mfer has joined #openstack-sdks20:26
*** mfer has quit IRC21:06
*** rmcall has quit IRC21:57
-openstackstatus- NOTICE: The openstack-infra/config repo will be frozen for project-configuration changes starting at 00:01 UTC. If you have a pending configuration change that has not merged or is not in the queue, please see us in #openstack-infra.21:58
*** rmcall has joined #openstack-sdks21:59
*** rmcall has quit IRC22:03
*** rmcall_ has joined #openstack-sdks22:04
*** sigmavirus24 is now known as sigmavirus24_awa22:15
*** briancurtin has quit IRC22:30
*** bknudson has quit IRC22:36
*** openstackgerrit has quit IRC22:47
*** openstackgerrit_ has joined #openstack-sdks22:47
*** openstackgerrit_ is now known as openstackgerrit22:48
*** briancurtin has joined #openstack-sdks23:05
*** terrylhowe has quit IRC23:14
*** rmcall_ has quit IRC23:33
*** rmcall has joined #openstack-sdks23:35
*** bknudson has joined #openstack-sdks23:36
*** bknudson has quit IRC23:36
*** bknudson has joined #openstack-sdks23:39
*** openstackstatus has quit IRC23:55
*** openstack has joined #openstack-sdks23:56
*** bknudson has quit IRC23:56
*** bknudson has joined #openstack-sdks23:58

Generated by 2.14.0 by Marius Gedminas - find it at!