| -openstackstatus- NOTICE: A failing log storage endpoint has been removed, you can recheck any recent jobs with POST_FAILURE where logs have failed to upload | 04:44 | |
| *** hyunsikyang has joined #openstack-fenix | 08:09 | |
| *** tojuvone has quit IRC | 08:31 | |
| *** tojuvone has joined #openstack-fenix | 08:32 | |
| tojuvone | Hi hyunsikyang, sorry did no notice earlier | 09:07 |
|---|---|---|
| tojuvone | btw, there was a bug with haproxy as Fenix needed authentication even with version discovery. Now I have a fix and can test running Fenix while it is being switched over to another controller during upgrade | 09:29 |
| hyunsikyang | Oh Hi | 09:48 |
| hyunsikyang | Long time no see. | 09:48 |
| hyunsikyang | good!! | 09:49 |
| hyunsikyang | And I have a something to tell you. | 09:49 |
| hyunsikyang | Tacker had a some problem so we almost fixed. So, tacker-fenix will be merged soon! | 09:49 |
| tojuvone | Yes, I saw. Just rebase hopefully | 10:18 |
| hyunsikyang | And our presentation was not going summit. | 10:58 |
| hyunsikyang | So sad.. | 10:58 |
| hyunsikyang | BTW, Why don't we think about next item for Fenix? | 10:58 |
| tojuvone | Well, now I am kind of doing in real env instead of DevStack. That means running real upgrade with 3 controller, x storage and y compute nodes. In Fenix that might be just fixing couple of bugs found during that exercise. | 11:08 |
| tojuvone | Surely there are things missing, but there is stories in storyboard too | 11:09 |
| tojuvone | one should support cofiguration generations. would be nice to have ci. unit testing should be finalized | 11:10 |
| tojuvone | Surely one can think of different scenarios with example workflows. Like supporting nova-compute on controller nodes. | 11:12 |
| tojuvone | Real action plugin implementations I have kept outside Fenix scope. In real use and in demos, one would have action plugins to make backup, update os, update esw, updated controller, storage node,... | 11:14 |
| tojuvone | That was not hard when did demo on Rocky to Stein upgrade. Fenix workflow gives you the framework and flow, then simple action plugin just to make your wanted action in your kind of deployment target | 11:15 |
| tojuvone | Is it then like, DevStack target, TripleO, Kubernetes... as an example | 11:16 |
| hyunsikyang | right. Actually we are thinking about fenix plugin with kubernetes. | 11:34 |
| hyunsikyang | I am not sure where we start now. I think we can make a architecture first with kubernetes function such as evicition or pod disruption.. | 11:35 |
| tojuvone | ok, do you mean in tacker or new one in Fenix side? Currently Fenix has the one example that expects keystone authentication. | 11:36 |
| tojuvone | Anyhow that would be great :) | 11:39 |
| tojuvone | Oh and yes the CFP. I guess something like full blown ha rolling upgrade from Train to Ussuri would have been something, but it would have required a lot of work just for sake of demo to an event | 11:41 |
| hyunsikyang | I mean that I am thinking about how we adapt fenix to the kubernetes. | 11:42 |
| hyunsikyang | If we are doing this, we can show the maintenance function for container and vm ! | 11:43 |
| hyunsikyang | I just want to ask you about a plan :) | 11:43 |
| tojuvone | Well it is done once, just with limitation of same AODH interface and keystone. I have not given a thought of it more. | 11:47 |
| tojuvone | So there could be different way to notify VNFM side in K8S and different authentication. I think Fenix use case for Fenix should be to support Hybrid env with both OpenStack and Kubernetes. Would be a different thing to make things accessible trough kubectl | 11:55 |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!