*** tetsuro has joined #openstack-placement | 01:53 | |
*** tetsuro has quit IRC | 02:13 | |
*** tetsuro has joined #openstack-placement | 03:01 | |
*** tetsuro has quit IRC | 03:05 | |
*** tetsuro has joined #openstack-placement | 03:49 | |
*** tetsuro has quit IRC | 04:22 | |
*** tetsuro has joined #openstack-placement | 04:58 | |
*** takashin has joined #openstack-placement | 06:32 | |
*** tetsuro has quit IRC | 06:35 | |
*** ttsiouts has joined #openstack-placement | 07:15 | |
*** tssurya has joined #openstack-placement | 07:29 | |
*** helenafm has joined #openstack-placement | 07:43 | |
*** ttsiouts has quit IRC | 07:46 | |
*** ttsiouts has joined #openstack-placement | 07:46 | |
*** ttsiouts has quit IRC | 07:51 | |
*** ttsiouts has joined #openstack-placement | 08:04 | |
*** tssurya_ has joined #openstack-placement | 08:41 | |
*** tssurya has quit IRC | 08:41 | |
*** e0ne has joined #openstack-placement | 08:47 | |
*** takashin has left #openstack-placement | 09:00 | |
*** helenafm has quit IRC | 09:07 | |
*** helenafm has joined #openstack-placement | 09:08 | |
*** ttsiouts has quit IRC | 09:29 | |
*** ttsiouts has joined #openstack-placement | 09:29 | |
*** cdent has joined #openstack-placement | 09:31 | |
*** ttsiouts_ has joined #openstack-placement | 09:33 | |
*** ttsiouts has quit IRC | 09:34 | |
*** cdent has quit IRC | 09:50 | |
*** ttsiouts_ has quit IRC | 09:53 | |
*** ttsiouts has joined #openstack-placement | 09:53 | |
*** ttsiouts_ has joined #openstack-placement | 09:57 | |
*** ttsiouts has quit IRC | 09:58 | |
*** e0ne has quit IRC | 10:00 | |
*** helenafm has quit IRC | 10:03 | |
*** helenafm has joined #openstack-placement | 10:07 | |
*** tssurya_ is now known as tssurya | 10:10 | |
*** cdent has joined #openstack-placement | 10:16 | |
*** ttsiouts_ has quit IRC | 10:17 | |
*** ttsiouts has joined #openstack-placement | 10:18 | |
*** ttsiouts has quit IRC | 10:23 | |
*** cdent has quit IRC | 10:23 | |
*** cdent has joined #openstack-placement | 10:26 | |
*** belmoreira has joined #openstack-placement | 10:45 | |
*** ttsiouts has joined #openstack-placement | 11:07 | |
openstackgerrit | Chris Dent proposed openstack/placement master: Make placement testing easier on osx https://review.opendev.org/671249 | 12:28 |
---|---|---|
*** belmoreira has quit IRC | 12:52 | |
*** ttsiouts has quit IRC | 12:54 | |
*** ttsiouts has joined #openstack-placement | 12:55 | |
openstackgerrit | Chris Dent proposed openstack/placement master: Use integrated-gate-placement zuul template https://review.opendev.org/671257 | 12:55 |
*** ttsiouts has quit IRC | 12:59 | |
*** ttsiouts has joined #openstack-placement | 13:15 | |
*** artom has quit IRC | 13:19 | |
*** belmoreira has joined #openstack-placement | 13:31 | |
*** belmoreira has quit IRC | 13:36 | |
sean-k-mooney | cdent: currently there is no sdk support for placement correct? and we dont have a python-placementclient either correct | 13:53 |
cdent | sean-k-mooney: you can create a raw session to the placement endpoint in the sdk | 13:53 |
sean-k-mooney | cdent: so today you would either use a keystone auth cleint or a generic rest client form the sdk | 13:53 |
sean-k-mooney | cdent: right that is what i assumed | 13:53 |
cdent | i would guess https://review.opendev.org/#/c/656023/ provides an example | 13:54 |
sean-k-mooney | i was ping directly asking about integrating placmenet in horizon | 13:56 |
sean-k-mooney | they were asking about reusing osc-placmenet but i was/am suggesting they use the sdk | 13:56 |
sean-k-mooney | and or create a placement module in the sdk instead of creating the fucntion in horizon | 13:56 |
cdent | yes, I would suggest using the sdk and create URLs, and are parses responses in the simple way | 13:56 |
sean-k-mooney | possibly by porting the nova code | 13:56 |
cdent | I'm not sure the nova code is sufficiently generic | 13:57 |
*** artom has joined #openstack-placement | 13:57 | |
sean-k-mooney | well im assuming all you would do in horizon is list the RPs | 13:57 |
cdent | which is why I've been resistant to creating a specific placement module in the skd: the URLs and the bodies are just URLs and JSON. Wrapping it in a bunch of a names and classes and other stuff is _less_ flexible and generic | 13:57 |
sean-k-mooney | maybe show allocations | 13:58 |
cdent | yeah, in which case some simple GETs is easy | 13:58 |
sean-k-mooney | ya i was thinking it would be nice to have a small wraper around get /resouce_providers | 13:58 |
sean-k-mooney | i could see wanting to eventully have all the endpoint but anyway i point them here but they have not joined so just an fyi that someone might ask about it later | 13:59 |
cdent | I guess i'm the wrong guy to be talking to about this: To me, adding a wrapper would get in the way of what i want: the dict that comes from the returned JSON. Some kind of class or whatnot is ... annoying. | 14:00 |
sean-k-mooney | cdent: i assume you agree using the raw client form the sdk is infintly better then embeding osc-placmenet in horizon | 14:00 |
cdent | But I realize that I'm often on my own when it comes to this sort of thing, so whatever people want we should do. I don't want to be roadblock. | 14:01 |
cdent | yes, definitely agree on that | 14:01 |
cdent | using osc-placemnet in horizon would be a _horrible_ idea | 14:01 |
sean-k-mooney | well i was thinking of a moduel with a much of function that just took the different qurey args as kwargs adn the required ones as postionals and return the dict of the raw reponce | 14:02 |
sean-k-mooney | e.g. a very light shim so you know what the parmater that endpoint accepts are | 14:02 |
sean-k-mooney | but anyway ill leave that as an exercise to the user :) | 14:03 |
*** e0ne has joined #openstack-placement | 14:05 | |
*** Bidwe_jay has joined #openstack-placement | 14:08 | |
*** altlogbot_3 has quit IRC | 14:09 | |
*** belmoreira has joined #openstack-placement | 14:10 | |
Bidwe_jay | hi all.....want to create osc-placement client in horizon | 14:11 |
Bidwe_jay | to get resource provider data if anyone have any idea it will be helpfull... | 14:11 |
cdent | Bidwe_jay: we just talked about that, let me get you a log link | 14:12 |
cdent | Bidwe_jay: http://eavesdrop.openstack.org/irclogs/%23openstack-placement/%23openstack-placement.2019-07-17.log.html#t2019-07-17T13:53:21 | 14:12 |
Bidwe_jay | Thank you! | 14:13 |
cdent | that's not fully up to date yet, it will take a few minutes to get to the last thing said which was [t 1BtM] | 14:13 |
purplerbot | <sean-k-mooney> but anyway ill leave that as an exercise to the user :) [2019-07-17 14:03:10.707724] [n 1BtM] | 14:13 |
sean-k-mooney | Bidwe_jay: its effectivly what we spoke about. e.g. your better off just using the sdk to get the raw placemnet rest client and doing direct GET requests for the info you want | 14:16 |
sean-k-mooney | Bidwe_jay: you could optionally add function to the sdk to construct the sdk | 14:16 |
sean-k-mooney | *urls | 14:16 |
*** irclogbot_0 has quit IRC | 14:17 | |
sean-k-mooney | and retrun the respoce dicts already deserialsed form json if you wantted to | 14:17 |
*** altlogbot_2 has joined #openstack-placement | 14:18 | |
*** belmoreira has quit IRC | 14:21 | |
Bidwe_jay | sean-k-mooney: ok | 14:21 |
*** altlogbot_2 has quit IRC | 14:22 | |
*** belmoreira has joined #openstack-placement | 14:23 | |
*** altlogbot_2 has joined #openstack-placement | 14:27 | |
*** altlogbot_2 has quit IRC | 14:32 | |
*** altlogbot_0 has joined #openstack-placement | 14:34 | |
*** altlogbot_0 has quit IRC | 14:38 | |
efried | sean-k-mooney, Bidwe_jay, cdent: Let me weigh in here: | 14:39 |
*** altlogbot_0 has joined #openstack-placement | 14:40 | |
efried | - Yes, the patch cdent points to (https://review.opendev.org/#/c/656023/) proves that you can use openstacksdk Connection to get a Proxy to the placement service and use raw get/put/etc. methods. | 14:40 |
efried | - No, there is no python-placementclient and there never will be. | 14:40 |
efried | - osc-placement is for CLI and nothing else. | 14:41 |
efried | - Support for other-than-raw-json wrappers around placement APIs, if such a thing is ever desired & developed, should go into the openstacksdk and follow their object modeling. | 14:42 |
*** artom has quit IRC | 14:42 | |
efried | - For now, if what you want is to do simple queries from horizon to placement, you could use either keystoneauth1 or openstacksdk to create your connection; but either way you would be using primitive get/put/etc. dealing with raw json. | 14:43 |
*** altlogbot_0 has quit IRC | 14:44 | |
*** altlogbot_0 has joined #openstack-placement | 14:46 | |
efried | Should you use ksa or sdk? I would have to understand a bit more about how horizon likes to communicate with the other services it talks to. I would say use sdk if that's possible. Does horizon use sdk for anything else? Does it know about clouds.yaml? And/or does it use oslo.config opts to configure connections to services? | 14:47 |
sean-k-mooney | efried: yep that more or less aligns with my preferences as well | 14:48 |
cdent | (just to throw a wrench in the works, for completeness, another option is do it all in javascript from the browser) | 14:48 |
sean-k-mooney | and ya the question re does horizon already use sdk vs keystone auth are valid and likely the main factor for chosing one over the other | 14:49 |
sean-k-mooney | cdent: you could but you would have to implement support for getting and handeling the keysone token | 14:49 |
*** altlogbot_0 has quit IRC | 14:50 | |
cdent | I was assuming that horizon already did that, but if not, then yeah. | 14:50 |
sean-k-mooney | but yes javascript woudl be an option but i guess it depends on what horizon does normally | 14:50 |
* cdent is not a horizon user | 14:50 | |
sean-k-mooney | i am a user somethimes but never really looked into how it works much | 14:50 |
*** altlogbot_3 has joined #openstack-placement | 14:52 | |
stephenfin | efried, cdent: Would you be open to adding a 'usage' column to 'openstack resource provider inventory list' and possibly deprecating 'openstack resource provider usage show' | 14:55 |
stephenfin | It feels odd that the former shows me everything about inventory for a given resource provider _except_ how much of it has been consumed | 14:55 |
*** altlogbot_3 has quit IRC | 14:56 | |
cdent | stephenfin: would you just do it, or do it with an arg? | 14:57 |
cdent | probably the latter as it adds some expense | 14:57 |
stephenfin | cdent: That depends. Is it (fetching usage) an expensive operation? | 14:57 |
stephenfin | Jinx | 14:57 |
cdent | it's not expensive in an of itself, but it does mean more requests | 14:58 |
*** altlogbot_3 has joined #openstack-placement | 14:58 | |
stephenfin | For a client tool, that doesn't seem _too_ bad? | 14:58 |
cdent | no, it doesn't | 14:58 |
cdent | I suppose the other concern would be whether the added arg would provide a guard against people doing results processing that doesn't expect the additional info | 14:59 |
edleafe | ll | 15:00 |
edleafe | doh! | 15:00 |
*** altlogbot_3 has quit IRC | 15:02 | |
efried | cdent, stephenfin: aren't we obligated to preserve some kind of backward compatibility of formatting when args don't change? | 15:02 |
efried | or only for tabular output? | 15:02 |
stephenfin | I haven't a clue, but I can ask dtroyer in a bit to find out | 15:02 |
cdent | stephenfin: if dtroyer likes the idea, my general rules is "sure, why not?" | 15:04 |
*** altlogbot_1 has joined #openstack-placement | 15:04 | |
* stephenfin likes that rule | 15:04 | |
stephenfin | (Except when it comes to very large, hard to maintain/rip out features :)) | 15:04 |
*** artom has joined #openstack-placement | 15:05 | |
*** altlogbot_1 has quit IRC | 15:08 | |
*** belmoreira has quit IRC | 15:08 | |
*** ttsiouts has quit IRC | 15:13 | |
*** ttsiouts has joined #openstack-placement | 15:14 | |
*** altlogbot_0 has joined #openstack-placement | 15:14 | |
*** altlogbot_0 has quit IRC | 15:18 | |
*** ttsiouts has quit IRC | 15:18 | |
*** tssurya has quit IRC | 15:19 | |
*** altlogbot_0 has joined #openstack-placement | 15:20 | |
*** altlogbot_0 has quit IRC | 15:24 | |
*** altlogbot_3 has joined #openstack-placement | 15:34 | |
*** altlogbot_3 has quit IRC | 15:38 | |
*** belmoreira has joined #openstack-placement | 15:41 | |
*** belmoreira has quit IRC | 15:42 | |
*** altlogbot_0 has joined #openstack-placement | 15:46 | |
*** altlogbot_0 has quit IRC | 15:50 | |
*** helenafm has quit IRC | 15:51 | |
*** altlogbot_1 has joined #openstack-placement | 15:52 | |
*** artom has quit IRC | 15:52 | |
*** altlogbot_1 has quit IRC | 15:57 | |
*** artom has joined #openstack-placement | 15:59 | |
*** e0ne has quit IRC | 16:00 | |
*** altlogbot_2 has joined #openstack-placement | 16:04 | |
*** altlogbot_2 has quit IRC | 16:08 | |
*** altlogbot_3 has joined #openstack-placement | 16:10 | |
*** altlogbot_3 has quit IRC | 16:14 | |
*** irclogbot_2 has joined #openstack-placement | 16:14 | |
*** altlogbot_1 has joined #openstack-placement | 16:15 | |
*** irclogbot_2 has quit IRC | 16:17 | |
*** efried is now known as efried_rollin | 16:21 | |
*** altlogbot_1 has quit IRC | 16:34 | |
*** altlogbot_3 has joined #openstack-placement | 16:36 | |
*** artom has quit IRC | 16:42 | |
*** artom has joined #openstack-placement | 17:00 | |
openstackgerrit | Chris Dent proposed openstack/placement master: Move rc_cache onto RequestContext https://review.opendev.org/671341 | 17:42 |
cdent | efried_rollin, gibi : that ^ seems like it ought to address the resource class cache problem and at least superficially it doesn't seem to hurt too much | 17:57 |
cdent | that's my EOD | 17:58 |
*** cdent has quit IRC | 17:58 | |
*** e0ne has joined #openstack-placement | 18:36 | |
*** e0ne has quit IRC | 18:52 | |
*** efried_rollin is now known as efried | 20:18 | |
*** artom has quit IRC | 20:18 | |
*** e0ne has joined #openstack-placement | 20:38 | |
*** e0ne has quit IRC | 21:15 | |
*** irclogbot_3 has joined #openstack-placement | 21:23 | |
*** artom has joined #openstack-placement | 23:56 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!