Earlier  
Later
New Search
Posted Nick Remark
#openstack-sdks - 2017-04-20
17:06:36 dtroyer_zz \o/
17:06:52 mordred doing that curently would screw some things, but if I can get generic discovery working consistently and overridably when needed, then we can get rid of them
17:07:31 mordred (which is one of the things that I think will wind up being invasive ksa-side - currently it'll handle doing discovery, but you have no idea what version it actually discovered)
17:09:04 openstackgerrit Merged openstack/cliff master: Add smart help formatter for command parser https://review.openstack.org/454613
17:17:51 stevemar dtroyer_zz: shouldn't be a big deal about missing that one, Rui's patch only refactored things right?
17:18:24 rukh ping openstack-sdks: Looking at this https://developer.openstack.org/sdks/python/openstacksdk/users/resources/orchestration/v1/resource.html#openstack.orchestration.v1.resource.Resource I was able to get the physical_resource_id for the "Controller" resource and I think must be additional resources associated with this physical id .. I see the class this inherits from has a list() function. but I don't think I can access it like this:
17:18:26 rukh resources = self.conn.orchestration.resources(stack)\
17:18:30 rukh resources = self.conn.orchestration.resources(stack)
17:19:30 rukh resource_id = resource.physical_resource_id
17:20:02 rukh node = resources.get(resource_id,
17:20:02 rukh Controller)
17:20:53 rukh is it possible to access the base class resource members ?
17:21:29 edleafe rukh: since this is Python, yes
17:21:59 edleafe rukh: Also, when posting more than a line or two of code, using a paste service helps make the code more readable
17:22:08 edleafe and doesn't clutter the channel
17:22:16 rukh edleafe, apologies..
17:22:18 openstackgerrit Dean Troyer proposed openstack/osc-lib master: Add token auth test https://review.openstack.org/458597
17:22:24 edleafe rukh: no worries
17:22:46 dtroyer_zz stevemar: can you look at ^^^ and the associated revert if you have a minute?
18:15:59 openstackgerrit Carlos Goncalves proposed openstack/python-openstacksdk master: Add data plane status support to Network Port obj https://review.openstack.org/458608
18:47:22 openstackgerrit Dean Troyer proposed openstack/python-openstackclient master: Correct auth issue in master os-client-config" https://review.openstack.org/458615
18:57:37 openstackgerrit Merged openstack/python-openstacksdk master: Add data plane status support to Network Port obj https://review.openstack.org/458608
19:20:38 openstackgerrit Carlos Goncalves proposed openstack/python-openstackclient master: Add 'data_plane_status' option to Port classes https://review.openstack.org/458626
20:11:37 openstackgerrit Merged openstack/keystoneauth master: Updated from global requirements https://review.openstack.org/455926
20:18:09 openstackgerrit Merged openstack/osc-lib master: Revert "Avoid to authenticate twice" https://review.openstack.org/458586
20:18:16 openstackgerrit Merged openstack/osc-lib master: Add token auth test https://review.openstack.org/458597
20:26:44 rukh ping openstack-sdk: Am a little stuck: https://pastebin.com/n21Ymxnf was able to get the resource_id for the Controller resource but having trouble getting further to get the nodes associated with that resource
20:27:22 briancurtin if you’re using heatclient you may be better off asking in #openstack-heat
20:29:28 rukh briancurtin, I am actually moving that code since the clients I read are deprecated to use the openstack sdk calls
20:29:58 briancurtin i don’t know that they’re actually deprecated. is that documented somewhere?
20:34:51 rukh I might have understood "access to legacy clients" as deprecated just like the individual cli clients are deprecated but
20:46:36 openstackgerrit Carlos Goncalves proposed openstack/python-openstackclient master: Add 'data_plane_status' option to Port classes https://review.openstack.org/458626
#openstack-sdks - 2017-04-21
00:53:11 chatter29 hey guys
00:53:14 chatter29 allah is doing
00:53:18 chatter29 sun is not doing allah is doing
00:53:22 chatter29 to accept Islam say that i bear witness that there is no deity worthy of worship except Allah and Muhammad peace be upon him is his slave and messenger
09:23:44 openstackgerrit David Rabel proposed openstack/python-openstackclient master: Make test_server.py more elegant and simple https://review.openstack.org/458785
09:38:02 openstackgerrit David Rabel proposed openstack/python-openstackclient master: Make test_server.py more elegant and simple https://review.openstack.org/458785
11:06:48 openstackgerrit David Rabel proposed openstack/python-openstackclient master: Make test_server.py more elegant and simple https://review.openstack.org/458785
11:43:26 mordred edleafe: what? you can't just casually read that giant thing? but discovery is so simple ...
12:05:22 cdent mordred: I haven't read that thing yet, and I'm not sure I can promise to get it today because it is sunny and warm for once and I feel like I may deserve a break today. I may seek out livestock to sit on.
12:06:32 dtroyer_zz cdent: take that break before officially assuming your new task list… congrats
12:07:02 cdent thanks dtroyer_zz
12:12:03 mordred cdent: you should always sit on livestock!
12:12:12 mordred cdent: and yes, also, congrats!
12:12:34 cdent "A chicken in every pot, a bull under every butt."
12:12:49 mordred cdent: if you can't get to it today, no worries. edleafe has suggested that he thinks it actually is appropriate to submit to the api-wg repo (with a little bit of genericizing)
12:13:09 mordred so I may just work on that a little bit before you get to it anyway
12:13:46 cdent ah, well, cool
12:14:05 cdent the api-wg shall grow to consume all guidelines on how be midly sane
12:14:12 mordred (I replied ot him that I think there are maybe three documents a) how discoverability _should_ work at a perfect point in the future b) how clients can properly consume discoverability today c) any additional things shade does on top of b
12:15:11 mordred and if we have b, we can highlight with a the value of services implementing the guidelines "look at the fewer calls / less work in the client algorithm if you do discoverability corectly"
12:15:56 mordred anywho - I'll work on an edit nad send it to your repo - also, de-shadifying it a little might help dtroyer_zz to tell me what's broken in it from his pov too
12:16:21 dtroyer_zz mordred: I'm about 3/4 through it, fwiw, and only two comments so far
12:16:59 cdent mordred: a and b sound very api-wg useful
12:22:45 cgoncalves can we get a new release of osc-lib (>1.4.0) so that we can get https://review.openstack.org/#/c/458586/ ? all patch sets are failing to verify with osc-lib==1.4.0
12:24:33 dtroyer_zz cgoncalves: I submitted it last night: https://review.openstack.org/458652 will ping release team again
12:25:10 mordred dtroyer_zz: woot!
12:25:33 mordred dtroyer_zz: oh, also, since you're making a release I'm guessing you're aware, but I also pinged you in response to a user human on the twitters
12:25:46 cgoncalves dtroyer_zz: ah, thanks!
12:25:56 dtroyer_zz ah, twitter… I should pay more attention to that
12:26:18 dtroyer_zz and yes, that's what osc-lib 1.5.0 fixes
12:27:10 mordred I figured
12:27:20 mordred dtroyer_zz: did you get bit by transitive dependency hell?
12:27:20 dtroyer_zz it also makes me want to focus purely on a static-linked client
12:27:38 mordred yah
12:27:50 dtroyer_zz no actually, it was an 'optimization' for auth that I didn't have a complete test for tokens
12:28:00 mordred oh. gross
12:28:10 dtroyer_zz I think I know why it passed the other tests and am working on that now
12:28:28 mordred I actually had a dream last night that yesterday's shade release had missed testing something and that I woke up to a ton of people being angry that I'd broken them
12:29:25 dtroyer_zz I have those with every 'git review' of the releases repo
12:29:45 dtroyer_zz and don't release o-c-c yet, that's also on my short list to fix
12:31:01 dtroyer_zz summary: osc-s -tips job is broken in unusual and mysterious ways that are not easily duplicated in my local environment
12:32:17 cgoncalves dtroyer_zz: +workflow! 1.5.0 is on the way :-)
12:33:12 mordred dtroyer_zz: yah - I looked at the job fail and couldn't figure out why
12:33:19 dtroyer_zz cgoncalves: \o/ it'll take the u-c update to merge to unblock the OSC gate, everything else should be happy with the release
12:33:21 mordred dtroyer_zz: but have I mentioned how happy I am we added that job?
12:34:15 dtroyer_zz I set that job to run OSC in debug mode finally and it's a wealth of failures, and for different reasons, mostly due to removing the ksc auth thing from o-c-c (that needed to go) but that exposed a totally different problem that we'd been masking
12:35:06 dtroyer_zz the osc-lib 1.4.0 just totally confused things yesterday
12:39:45 mordred _awesome_
12:48:38 openstackgerrit Stephen Finucane proposed openstack/python-openstackclient master: DNM For the love of FSM https://review.openstack.org/458819
13:39:27 openstackgerrit Carlos Goncalves proposed openstack/python-openstackclient master: Add 'data_plane_status' option to Port classes https://review.openstack.org/458626
13:49:59 stevemar dtroyer_zz: about https://review.openstack.org/#/c/444924/15/openstackclient/compute/v2/server.py
13:50:06 stevemar where are the new args being used?
13:50:13 stevemar did i miss something?
13:52:33 dtroyer_zz stevemar: new args? do you mean the two new options?
13:58:48 stevemar yes
14:16:49 dtroyer_zz they are just convenience options, I'm not sure what motivated the change. Similar to what I did in https://review.openstack.org/302795 to make users' lives simpler
14:17:53 cgoncalves dtroyer_zz: re https://review.openstack.org/#/c/458626/ is there any PUT-only argument in osc where osc does a POST+PUT (in a class Create)?
14:18:40 dtroyer_zz There are places where we do multiple calls to achieve the desired result, I'd have to look for some specific examples
14:19:12 dtroyer_zz remember, we're first providing a consistent user interface, and doing what is necessary on the backside to make up for the REST APIs
14:19:48 cgoncalves dtroyer_zz: don't bother if you don't remember by heart. I'll look up
14:21:05 cgoncalves understood. it will feel weird to me having multiple calls for this case in particular but okay
14:21:37 dtroyer_zz once you get past the 1-to-1 mapping of commands to REST APIs it gets easier :)
14:22:45 dtroyer_zz volume create does this
14:23:15 cgoncalves yeah, but in this case it doesn't make much since IMO because data_plane_status represents the status of the underlying data plane and when one creates a new port it doesn't know the status beforehand
14:23:17 dtroyer_zz v1 has 1 extra, v2 has 2 extra calls depending on options
14:24:06 cgoncalves ok, I'll have a look. it's not much of a technical matter but rather if it makes sense allowing setting it on port creation
14:24:16 dtroyer_zz ok, that's what I mean about the use case, if it really is never going to be used we don't need it.
14:24:37 dtroyer_zz but I don't know exactly how it is used…

Earlier   Later