| *** akhil_jain has joined #openstack-valence | 03:40 | |
| openstackgerrit | Anusha Ramineni proposed openstack/valence-specs master: Add vendor extensible framework https://review.openstack.org/469817 | 04:09 |
|---|---|---|
| openstackgerrit | Anusha Ramineni proposed openstack/valence master: Fix create_podmanager to store auth details https://review.openstack.org/470851 | 05:03 |
| openstackgerrit | Anusha Ramineni proposed openstack/valence-specs master: Add pooled resources management https://review.openstack.org/471322 | 10:07 |
| *** shuquan has joined #openstack-valence | 14:32 | |
| *** ananth_n has joined #openstack-valence | 14:54 | |
| *** hubian has joined #openstack-valence | 14:55 | |
| *** ramineni_ has joined #openstack-valence | 14:58 | |
| ananth_n | hello everyone | 14:59 |
| *** mrittika has joined #openstack-valence | 14:59 | |
| mkrai | Hi All | 14:59 |
| mrittika | hi | 14:59 |
| ramineni_ | hi | 14:59 |
| shuquan | hi | 14:59 |
| hubian | hi | 14:59 |
| ananth_n | there are 2 agenda items listed for today | 15:00 |
| *** jfding has joined #openstack-valence | 15:00 | |
| ananth_n | let's start with the multi podm support item | 15:00 |
| ananth_n | who posted it? | 15:01 |
| ananth_n | i mean, the agenda item :) | 15:01 |
| mrittika | i did | 15:01 |
| mrittika | wanted to know if there is a patch and can we build and test it | 15:02 |
| ananth_n | hi mrittika. did you have specific question on it? we have a BP for the multi podm scheduler. we need to write the spec | 15:02 |
| ananth_n | multi podm support is integrated. hubian wrote it | 15:02 |
| mrittika | who will be writing the spec | 15:02 |
| ananth_n | lin_yang has assigned the spec to himself. he is unable to join the meeting today. so I will follow up offline | 15:03 |
| ananth_n | *assigned the BP | 15:03 |
| jfding | I will tell him | 15:04 |
| ananth_n | thanks jfding | 15:04 |
| jfding | np :) | 15:04 |
| mrittika | ok. @hubian is there a simple sketch of the flow available? | 15:04 |
| hubian | the patch for the mutli podm is still blocked on the choice of the callable realation between podm and redfish | 15:05 |
| mrittika | ok. why is it blocked | 15:06 |
| ananth_n | @hubian, i'm sorry i didn't follow | 15:06 |
| mkrai | hubian: By callable relation you mean? | 15:06 |
| ananth_n | could you explain. i know you are working on one refactoring. but what is the callable relation? | 15:07 |
| *** ntpttr_laptop has joined #openstack-valence | 15:07 | |
| ananth_n | hi ntpttr_laptop | 15:07 |
| ntpttr_laptop | hi all, sorry to be a little late. IRC client was having issues so I'm on webchat | 15:07 |
| ntpttr_laptop | thanks ananth_n | 15:07 |
| hubian | My thought of support for.the mutli podm needs.to do refactor of our codes on the redfish.py and make podmanager instance function to response request.from user | 15:07 |
| ntpttr_laptop | hubian: I think that makes sense to me, yes | 15:08 |
| mrittika | ok do you need help? | 15:08 |
| ntpttr_laptop | and isn't that needed for your spec for multiple southbound drivers ramineni? | 15:09 |
| ananth_n | yes, that seems correct. but i'm afraid i i didn't understand the issue you are facing | 15:09 |
| ramineni_ | ntpttr_laptop: yes | 15:09 |
| hubian | eh for example , we implement conpose in redfish.py | 15:09 |
| ramineni_ | hubian: can we keep all the node specific operations in redfish? | 15:09 |
| ramineni_ | as it is there already now | 15:09 |
| ntpttr_laptop | hubian: you're suggesting moving the compose function out of redfish.py? But it's a redfish call | 15:10 |
| ntpttr_laptop | or did I misunderstand | 15:10 |
| hubian | yeah , but i also thought raminani's idea is also make sense | 15:10 |
| hubian | i thought your both are ok , just need a finialize decision | 15:11 |
| hubian | create a property called redfish for every podmanager instance is ok | 15:12 |
| ntpttr_laptop | hubian: yeah I think it would probably be good to keep node operations and redfish calls in redfish.py, and then ramineni_ can implement her new driver to handle those same calls | 15:12 |
| ananth_n | +1 | 15:12 |
| ramineni_ | ntpttr_laptop: right | 15:13 |
| ntpttr_laptop | hubian: sure, so would you move all the redfish functions into the pod manager layer? | 15:13 |
| hubian | ok. so this could be the final answer | 15:13 |
| ntpttr_laptop | they need to be abstracted somehow so that more than just redfish can be called, so maybe they can stay in their own files but podm can create instances of them when requested | 15:13 |
| ntpttr_laptop | hubian: sure, makes sense to me | 15:13 |
| ramineni_ | hubian: having property of redfish in podmanager and in controller, instance of podmanager calling function on redfish | 15:14 |
| hubian | yes | 15:14 |
| ramineni_ | hubian: this is what my understanding , just making sure everyone are on same page, | 15:14 |
| hubian | ok | 15:14 |
| ramineni_ | hubian: thanks | 15:14 |
| ananth_n | since we don't say no to non-redfish drivers, which approach would be a better design? | 15:15 |
| ananth_n | folks can 'vote' and we'll have a decision :) | 15:16 |
| hubian | if so , we just need to figure out which functions need to be in redfish like compose and decompose maybe also have some others | 15:16 |
| ntpttr_laptop | hubian: what about all the 'get' requests to view resources? | 15:16 |
| hubian | i refer to make these get functions in podmanager instance functions | 15:17 |
| ntpttr_laptop | hubian: okay, so how will you handle multiple drivers besides redfish then? | 15:18 |
| hubian | actually we do to get always we have need to know which podm | 15:18 |
| mrittika | you mean multiple redfish or multiple PODM | 15:18 |
| ntpttr_laptop | yes that's true, but I still think the podm should be making calls out to drivers to make the calls | 15:19 |
| ntpttr_laptop | like in podm store which driver to use, and if redfish then podm makes the redfish.get_systems call | 15:19 |
| ntpttr_laptop | but get_systems is still in redfish.py because it's a redfish call | 15:19 |
| hubian | yeah | 15:19 |
| ntpttr_laptop | the podm provides all the redfish instance connection info | 15:20 |
| ananth_n | we have to consider possibility multiple versions of the podm. | 15:20 |
| ntpttr_laptop | to me, that makes the most sense | 15:20 |
| ntpttr_laptop | ananth_n: yes, that can be stored in podm and maybe we can have different redfish drivers for different versions | 15:20 |
| ntpttr_laptop | a 1.2 and a 2.X or something | 15:20 |
| hubian | your are right | 15:20 |
| ntpttr_laptop | okay that's making sense I think | 15:21 |
| ananth_n | if i understood correctly, we have an instance of a podm object which will abstract all podm operations. correct? | 15:21 |
| hubian | yeah | 15:21 |
| ananth_n | so for multi podm, we create multiple such objects | 15:21 |
| ntpttr_laptop | ananth_n: I think the podm instance is a collection of info stored in the database about what kind of calls to make and the connection info needed | 15:21 |
| ntpttr_laptop | was there anything else we wanted to go over? Looks like 7 more minutes | 15:23 |
| ananth_n | ok, looks like i lost the train of thought somewhere. woill follow up over email | 15:23 |
| ntpttr_laptop | ananth_n: if we don't have anything else, I'd be okay making sure we're all on the same page with this | 15:24 |
| ntpttr_laptop | what's not lining up do you think? | 15:24 |
| hubian | i will send a detail email to describe the mutli podm which i shouls. | 15:24 |
| ntpttr_laptop | that sounds good hubian, thanks | 15:25 |
| ananth_n | thanks hubian | 15:25 |
| hubian | which i should have finished last week , we can detail talk about and finialize it | 15:25 |
| ananth_n | ntpttr_laptop let me read thru' the chat agin. things might clear up | 15:26 |
| hubian | :) | 15:26 |
| ananth_n | i also suggest that going fwd, we extend to 1 hour. this came up last week also | 15:26 |
| mrittika | ok.@ ananth_n are we going ot be able to talk about the task and queue mgr next week? | 15:26 |
| ntpttr_laptop | ananth_n: yes, we'd have to do something about that meeting we have scheduled right after this | 15:27 |
| ntpttr_laptop | but I agree we need 1 hour | 15:27 |
| ananth_n | if there's agreement on meeting 30 minutes earlier, i will update the wiki page | 15:27 |
| mrittika | no let's extend 30 later | 15:27 |
| ntpttr_laptop | I could do that if needed | 15:27 |
| ntpttr_laptop | mrittika: we have a meeting right after this | 15:27 |
| mrittika | so we start the same time and extend | 15:27 |
| ramineni_ | early +1 :) | 15:27 |
| ntpttr_laptop | mrittika: we can try and get it moved though | 15:27 |
| mrittika | that is only this week | 15:27 |
| ntpttr_laptop | oh good | 15:28 |
| shuquan | early +1 | 15:28 |
| ntpttr_laptop | didn't remember that | 15:28 |
| ananth_n | @mrittika 30 mins later might be difficult for hubian & shuquan | 15:28 |
| mrittika | for folks is US what works better? | 15:28 |
| mrittika | oh ok | 15:28 |
| ntpttr_laptop | I could do earlier. It would be 7:30am but that's not so bad | 15:28 |
| mrittika | ok. that's good then | 15:28 |
| ananth_n | 2:30 pm utc | 15:28 |
| hubian | early +1 | 15:28 |
| mrittika | early +1 | 15:28 |
| mrittika | bye then! | 15:29 |
| ananth_n | bye folks | 15:29 |
| *** mrittika has quit IRC | 15:29 | |
| ramineni_ | thanks all, bye | 15:29 |
| ntpttr_laptop | thank you all | 15:30 |
| hubian | see you | 15:30 |
| shuquan | 88 | 15:30 |
| *** hubian has quit IRC | 15:30 | |
| *** shuquan has quit IRC | 15:30 | |
| *** ananth_n has quit IRC | 15:32 | |
| *** ramineni_ has quit IRC | 15:48 | |
| *** ntpttr_laptop has quit IRC | 16:19 | |
| *** jfding has quit IRC | 17:50 | |
| -openstackstatus- NOTICE: The Gerrit service on review.openstack.org is being restarted now to clear some excessive connection counts while we debug the intermittent request failures reported over the past few minutes | 20:06 | |
| openstackgerrit | Nate Potter proposed openstack/valence master: Update dependencies in install script https://review.openstack.org/471920 | 21:11 |
| *** openstack has joined #openstack-valence | 23:15 | |
| *** ramineni_ has joined #openstack-valence | 23:57 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!