openstackgerrit | Merged openstack/python-valenceclient master: Add unittest for exceptions.py https://review.openstack.org/438843 | 00:46 |
---|---|---|
*** lin_yang has joined #openstack-valence | 00:49 | |
*** lin_yang has quit IRC | 08:17 | |
*** openstackgerrit has quit IRC | 10:18 | |
*** dasanind has quit IRC | 14:40 | |
*** dasanind has joined #openstack-valence | 14:41 | |
*** mkrai_ has joined #openstack-valence | 14:48 | |
*** lin_yang has joined #openstack-valence | 14:55 | |
*** shuquan has joined #openstack-valence | 14:59 | |
*** ananth_n has joined #openstack-valence | 15:01 | |
ananth_n | hi all | 15:02 |
shuquan | hi | 15:02 |
mkrai_ | Hi | 15:02 |
lin_yang | hello everyone | 15:02 |
ananth_n | so we have one talk accepted for Boston summmit and on as alternate | 15:02 |
ananth_n | so congrats to lin, shuquan *applause* | 15:03 |
ananth_n | :) | 15:03 |
mkrai_ | Congrats!! | 15:03 |
*** HuBian has joined #openstack-valence | 15:03 | |
HuBian | Hello all | 15:03 |
lin_yang | shuquan: thanks a lot for submitting that topic | 15:03 |
mkrai_ | I have a topic to discuss related to Keystone | 15:03 |
ananth_n | mkrai_ had a question on keystone integration. maybe we could discuss that quickly | 15:03 |
shuquan | you're welcome | 15:04 |
mkrai_ | Yes | 15:04 |
lin_yang | sure | 15:04 |
ananth_n | and if @ramineni is there, then Nate had asked for some details on the generic driver interface design. | 15:04 |
ananth_n | @mkrai floor is yours | 15:04 |
mkrai_ | Ok is there any plan of supporting keystone? | 15:05 |
mkrai_ | Currently there is no authentication for valence api service | 15:05 |
lin_yang | we should support keystone, which is the first choice for authientication | 15:06 |
mkrai_ | I am working on adding a new node API /nodes/{ID}/register to register a composed node to ironic | 15:06 |
mkrai_ | For that I would need the keystone authentication | 15:06 |
ananth_n | since we have multi-cloud support in the plan, and a different cloud orchestrator could use some other auth mechanism, tightly coupling valence with keystone is something I am not sure about | 15:06 |
mkrai_ | ananth_n: Everyone Openstack service is using keystone service | 15:06 |
mkrai_ | So I think that should not be an issue | 15:07 |
ananth_n | by multi cloud, i also mean different orchestrators | 15:07 |
ananth_n | i.e valence can work with kube, vmware, azure..... (big vision :) ) | 15:07 |
lin_yang | agree, we should make the authentication module configurable to allow choose different authentication module | 15:07 |
mkrai_ | ananth_n: Yes that will be the case. But for the ability to talk to ironic, keystone can be made default | 15:08 |
lin_yang | make it as middleware, if the cloud use keystone, admin can enable keystone middleware before valence api | 15:08 |
mkrai_ | lin_yang: Yes that's the plan to add it as a middleware | 15:09 |
mkrai_ | Ok I think I will register a bp for it | 15:11 |
mkrai_ | And then write the discussion over there | 15:11 |
ananth_n | my suggestion is to keep keystone bits within modules related to openstack. so the bigger valence codebase is agnostic of the auth mechanims | 15:11 |
ananth_n | *mechanism | 15:11 |
ananth_n | i.e. in this case of ironic integration, that specific module will auth with keystone (of the cloud to which new node must be added) and | 15:12 |
ananth_n | register the node | 15:12 |
ananth_n | we will not have a Valence level auth token | 15:12 |
ananth_n | because then we have to maintain one token for each cloud | 15:12 |
ananth_n | and different types of tokens for different orchestrators | 15:12 |
lin_yang | mkrai: when register node to ironic, it have to talk to keystone, at the same time, valence don't have to tight to keystone for it owned user authentication | 15:12 |
lin_yang | ananth_n: agree | 15:13 |
HuBian | yeah ~ i'm just think the same thought ~ | 15:13 |
mkrai_ | lin_yang: ananth_n agree | 15:13 |
mkrai_ | So I am going ahead with that design only | 15:13 |
mkrai_ | for the ironic communicate couple it with keystone | 15:14 |
ananth_n | mkrai_ thanks! will wait for BP | 15:14 |
mkrai_ | And leave the valence level auth for now | 15:14 |
mkrai_ | Is that fine? | 15:14 |
ananth_n | fine for me. let others also chip in | 15:14 |
lin_yang | it's fine to me | 15:15 |
lin_yang | mkrai: thanks, wait for bp | 15:16 |
HuBian | Shall we support different type of authentication at the same time ? Will this user case happend ? | 15:16 |
mkrai_ | The authentication part is always configurable as middleware | 15:17 |
mkrai_ | So I think we need not to work about it | 15:17 |
mkrai_ | worry* | 15:17 |
ananth_n | HuBian yes, different type of auth is possible. If I have a rack with one Openstack cloud and one VMWare cloud (for example), then each one might use a different auth mechanism | 15:18 |
lin_yang | HuBian: from my point of view, it will happen | 15:18 |
ananth_n | I haven't worked with VMWare, but just an example | 15:18 |
lin_yang | use valence to manage a datacenter, and it deploy multi-cloud on it | 15:19 |
ananth_n | mkrai's design supports configurablility | 15:19 |
HuBian | ok , wait for the BP , maybe I could get the answers from it ~ | 15:20 |
lin_yang | but in this case, valence might need self-owned authentication method | 15:20 |
mkrai_ | Yes I will try to explain the design on white board | 15:20 |
mkrai_ | Ok so I wanted to check on valenceclient status also | 15:21 |
mkrai_ | Is Jinxing available now? | 15:22 |
lin_yang | mkrai: sounds good. Jinxing submitted several modules for client, please align with him | 15:22 |
mkrai_ | Yes I saw that. Those are helper code | 15:23 |
shuquan | he seems not here right now. | 15:23 |
lin_yang | already 23mins, any other topics? | 15:23 |
mkrai_ | Ok | 15:24 |
*** HuBian has quit IRC | 15:24 | |
*** Jinxing has joined #openstack-valence | 15:24 | |
lin_yang | just let you know, I am working on sushy library this week to make sure it can operate composed node in RSD | 15:25 |
ananth_n | I don't have any other topics. There was one for Anusha, but I guess she isn't here today | 15:25 |
lin_yang | the composed node api is a bit different with redfish standard system | 15:25 |
mkrai_ | lin_yang: That's great | 15:25 |
lin_yang | okay | 15:25 |
Jinxing | I have something to talk about | 15:25 |
ananth_n | hi Jinxing | 15:25 |
lin_yang | Jinxing is here, go ahead | 15:25 |
mkrai_ | lin_yang: FYI, i have registered a bp for same | 15:26 |
mkrai_ | #link https://blueprints.launchpad.net/openstack-valence/+spec/sushy-support | 15:26 |
mkrai_ | hi Jinxing | 15:26 |
Jinxing | now my valenceclient console command is such as “valence node-list”, but the valence service start command is also is “valence start” | 15:26 |
HuBian_ | /join openstack-valence | 15:27 |
mkrai_ | Jinxing: I think it should be valence service-start | 15:27 |
*** HuBian_ has joined #openstack-valence | 15:28 | |
Jinxing | I think we should define which one to use the valence | 15:28 |
Jinxing | yes, the client and server are start with “valence” | 15:28 |
HuBian_ | seems i got a network problem just now ~ | 15:28 |
mkrai_ | Jinxing: Ok I got your point | 15:29 |
mkrai_ | Jinxing: We can rename the valence service to "valence-api" | 15:29 |
mkrai_ | Same as other OS projects | 15:29 |
Jinxing | it confilict when you try to install valence client after the valence server have been installed | 15:29 |
mkrai_ | Jinxing: agree. How about renaming the valence service to "valence-api"? | 15:30 |
Jinxing | This is what i want ot discuss. Chagne the valence service or the valence client console command prefix | 15:32 |
ananth_n | I am ok with that change | 15:32 |
ananth_n | so we won't have conflict with valence client command | 15:32 |
shuquan | +1 | 15:32 |
ananth_n | +1 | 15:32 |
lin_yang | +1 | 15:32 |
Jinxing | Although change the valence service console command prefix start with “valence-api”, but i feel it strange.Shall we change the valence service to like other project ,such as nova, | 15:34 |
shuquan | sorry guys, i have to leave. see you next week. | 15:34 |
*** shuquan has quit IRC | 15:34 | |
mkrai_ | Jinxing: Change the valence service to valence-api not the client | 15:35 |
Jinxing | Ok, I got it | 15:35 |
mkrai_ | Great!! | 15:35 |
Jinxing | So there should be a patch in valence service to change valence service start command, right? | 15:35 |
mkrai_ | Yes | 15:36 |
Jinxing | 👌 | 15:36 |
ananth_n | i need to drop too. | 15:37 |
ananth_n | catch you folks next week | 15:38 |
ananth_n | bye | 15:38 |
*** ananth_n has left #openstack-valence | 15:38 | |
lin_yang | thanks everyone | 15:38 |
mkrai_ | Bye! | 15:38 |
lin_yang | bye | 15:38 |
HuBian_ | bye | 15:40 |
*** HuBian_ has quit IRC | 15:40 | |
*** Jinxing has quit IRC | 15:47 | |
*** mkrai_ has quit IRC | 17:39 | |
*** lin_yang has quit IRC | 22:57 | |
*** ramineni_ has joined #openstack-valence | 23:56 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!