| *** hyunsikyang__ has quit IRC | 01:53 | |
| *** hyunsikyang has joined #openstack-fenix | 01:53 | |
| *** tojuvone has quit IRC | 03:05 | |
| *** tojuvone has joined #openstack-fenix | 03:06 | |
| *** tojuvone has quit IRC | 03:26 | |
| *** tojuvone has joined #openstack-fenix | 03:26 | |
| hyunsikyang | Hi tojuvon, when you have a time, please let me know:) | 04:28 |
|---|---|---|
| *** tojuvone has quit IRC | 05:15 | |
| *** tojuvone has joined #openstack-fenix | 05:16 | |
| tojuvone | Hi hyunsikyang, Now I am here. Yesterday I had a standalone K8S modified "Doctor VNFM" running to test new workflow. Now more or less just missing Fenix K8S workflow modifications for the AODH alarming to test the whole. Today I need to prepare "Fenix with ETSI FEAT03" presentation in OPNFV Closed Loop WG. | 05:43 |
| hyunsikyang | Oh good. when it start? is it regular meeting? | 05:44 |
| tojuvone | If arranged: Bi-weekly Thursdays at 7AM to 8AM Pacific Standard Time. | 05:49 |
| tojuvone | I could try to make these last bits to test this new "K8S workflow POC" and share the code and how to test. Will be simpler as no Doctor needed. Just need DevStack to bring Fenix, AODH, Keystone (and providing demo user/project used in testing) and then K8S cluster with 3 worker nodes. | 06:02 |
| hyunsikyang | good:) | 06:02 |
| hyunsikyang | If I can't join that meeting, could you share the slide for me? | 06:03 |
| tojuvone | I think it makes discussing all that further easier ;) | 06:03 |
| hyunsikyang | yes. I think so. | 06:03 |
| tojuvone | Yes, the meeting must be at bad time. I will surely share that material. | 06:04 |
| hyunsikyang | yes. midnight for me kk | 06:04 |
| hyunsikyang | thanks | 06:04 |
| tojuvone | Also not sure if anybody able to join (out of OPNFV), but this is link to meeting if that is possible https://zoom.us/j/5014627785 | 06:05 |
| hyunsikyang | I have a account for that. last time I joined OPNFV infra WG meeting | 06:05 |
| hyunsikyang | ) | 06:05 |
| tojuvone | Now I should make the material :o | 06:06 |
| tojuvone | I have some power point, but not happy with it and it does not mention K8S yet | 06:06 |
| hyunsikyang | could you give me a 5 min? | 06:06 |
| tojuvone | yes, sure | 06:06 |
| hyunsikyang | About the tacker integration, after alarming, | 06:07 |
| tojuvone | yes | 06:08 |
| hyunsikyang | fenix start maintenance using default workflow. right? such as maintenance or in scale ... | 06:08 |
| hyunsikyang | And tacker-plunin received these message and forward or act something and return the result to fenix. | 06:09 |
| hyunsikyang | act something with tacker and return. | 06:09 |
| tojuvone | There is currently default and vnf workflow. Yes, that is about what should be there in Tacker side. | 06:11 |
| tojuvone | Problem might be that what is the relation to VNF to make those actions and reply | 06:12 |
| tojuvone | Easier with ETSI FEAT03, as you could at least have some VNFD to define statically the behavior | 06:13 |
| hyunsikyang | https://review.opendev.org/#/c/643242/17/specs/train/vnf-rolling-upgrade.rst | 06:14 |
| hyunsikyang | so with this spec, I just want to know what is the scope of this implementation. | 06:14 |
| hyunsikyang | we have a two procedure for fenix in this spec. | 06:14 |
| hyunsikyang | So Do we develop real tacker-plugin function to manage the all of the message from fenix. | 06:15 |
| hyunsikyang | ? | 06:15 |
| hyunsikyang | or | 06:16 |
| hyunsikyang | tacker-plugin just can reply response message to fenix, when it received message such as PREPARE_MAINTENANCE, ADMIN_ACTION_DONE- | 06:16 |
| tojuvone | looking the spec.. | 06:16 |
| hyunsikyang | But | 06:17 |
| hyunsikyang | if you are busy today, | 06:17 |
| hyunsikyang | you don't need to use a time for it now. | 06:17 |
| hyunsikyang | I am just asking ..:) | 06:18 |
| hyunsikyang | now we developed code until alarming and return the message to fenix. | 06:18 |
| hyunsikyang | to working this procedure, and show the real maintance, we make that code too. but, some message does not supported by tacker now.. so | 06:19 |
| tojuvone | yes, so maybe this alarming this could be finalized if it works and then other aptch set to deal with those actions and reply | 06:19 |
| hyunsikyang | I just want to discuss with you. | 06:19 |
| tojuvone | so the "second" workflow in spec might be something to start with if scaling is not that easy to make. | 06:20 |
| tojuvone | PREPARE_MAINTENANCE and PLANNED_MAINTENANCE are very similar. Those could be implemented with ACK to Fenix | 06:22 |
| hyunsikyang | yes. So, one of the good way is that we can make a real demo with this. even the procedure is so simple. | 06:22 |
| tojuvone | Every time Fenix needs to make migration/live migration and it is done, there will be ADMIN_ACTION_DONE. You do not reply to this | 06:23 |
| hyunsikyang | Could you let me know where is the definition of each message? It needs to make a function | 06:23 |
| hyunsikyang | yes thesething should be clear.. | 06:23 |
| tojuvone | Then you surely need to process MAINTENANCE and MAINTENANCE_COMPLETE with ACK to Fenix | 06:24 |
| tojuvone | Let's see the Fenix documentation... | 06:25 |
| hyunsikyang | sorry to bother you.. | 06:25 |
| hyunsikyang | :(I know you are busy | 06:25 |
| tojuvone | No problem.. I gather here some links and let's see if that is enough to define "each message" | 06:27 |
| tojuvone | https://fenix.readthedocs.io/en/latest/user/baseworkflow.html | 06:27 |
| tojuvone | https://fenix.readthedocs.io/en/latest/user/notifications.html#project | 06:28 |
| tojuvone | So each notification/event will have the same message, but not all those variables. There is no specification for single state message content. Only what is described in that document | 06:30 |
| tojuvone | Also for the VNFM reply, there is only this: https://fenix.readthedocs.io/en/latest/api-ref/v1/index.html#id6 | 06:32 |
| tojuvone | so the " PUT /v1/maintenance/{session_id}/{project_id}/" | 06:32 |
| tojuvone | Information is there, but not see easy to handle as it is not by each state (MAINTENANCE, PLANNED_MAINTENANCE...) | 06:34 |
| tojuvone | One straight forward thing might be to run the Doctor test and look the Doctor app_manager/sample.py code to see that content | 06:35 |
| tojuvone | Running the Doctor test will print each message/event | 06:36 |
| tojuvone | code is here: https://github.com/opnfv/doctor/blob/master/doctor_tests/app_manager/sample.py | 06:36 |
| hyunsikyang | okie. I will look at it. | 06:37 |
| tojuvone | Maybe can find some log... I think I shared some of that, or Doctor CI run might have it | 06:37 |
| hyunsikyang | but tacker use a heat-stack so we should modify the code. | 06:38 |
| tojuvone | Surely best thing would be the document, but do nto have that | 06:38 |
| tojuvone | Doctor also make those actions relevant to heat stack via heat (scaling actions), but thing like Switch-over is not done via that. Don't know if even possible / feasible | 06:40 |
| hyunsikyang | as a perspective of takcer, we can finish the code until alarming. And then we can make a real scenario with procedure two in the spec. simple thing | 06:40 |
| hyunsikyang | I see. we need more discuss about it. | 06:41 |
| tojuvone | yes, I think that sound like a good plan | 06:41 |
| hyunsikyang | we will try to finish it first. And think the real demo! | 06:42 |
| hyunsikyang | really thank you for today discussion:) | 06:42 |
| tojuvone | hyunsikyang: Thank you for making all this :) That's great! | 06:45 |
| tojuvone | Old Doctor CI results seems to be wiped out.. But I might have some saved elsewhere.. | 06:46 |
| tojuvone | Here is some logs with the new VNFM, but they should give you mostly the message contents for each state.. | 06:54 |
| tojuvone | OPNFV Doctor test run: http://paste.openstack.org/show/787774/ | 06:54 |
| tojuvone | Fenix logging for Doctor test: http://paste.openstack.org/show/787775/ | 06:54 |
| tojuvone | In first one you can see what kind of message the VNFM receives (VNFM received data) and what it does and then reply (VNFM reply) | 06:56 |
| tojuvone | on the second link with Fenix log you can find similar (Sending "maintenance.planned" to project). For reply there is not the message shown, just project specifically if it is gotten (498677d1-e9e2-4930-ac05-965aa8d30858 in: ACK_PLANNED_MAINTENANCE). NOTE! As this is with new VNFM and nfv workflow, things are mostly just different in a way that each instance has separate message so they can be handler parallel. Meaning 'instance_ids' list only has | 07:08 |
| tojuvone | one instance always and not all the instances in current compute. Do not have to care about that. Also in log you can see some TIMEOUT AND migration retry as this env has not been the most stable one for AODH alarm delivery and alsoas Nova is pretty bad in doing many migrations in short time window to same compute hosts. | 07:08 |
| tojuvone | That is what I can find for now, just ask anytime if some doubt and I will reply when online :) | 07:10 |
| *** tojuvone has quit IRC | 08:06 | |
| *** tojuvone has joined #openstack-fenix | 08:06 | |
| *** tojuvone has quit IRC | 08:18 | |
| *** tojuvone has joined #openstack-fenix | 08:18 | |
| *** tojuvone has quit IRC | 08:30 | |
| *** tojuvone has joined #openstack-fenix | 08:30 | |
| *** tojuvone has quit IRC | 08:55 | |
| *** tojuvone has joined #openstack-fenix | 08:55 | |
| *** tojuvone has quit IRC | 09:41 | |
| *** tojuvone has joined #openstack-fenix | 09:41 | |
| *** openstackstatus has quit IRC | 09:45 | |
| *** tojuvone has quit IRC | 10:22 | |
| *** tojuvone has joined #openstack-fenix | 10:22 | |
| *** tojuvone has quit IRC | 10:23 | |
| *** tojuvone has joined #openstack-fenix | 10:24 | |
| *** tojuvone has quit IRC | 10:36 | |
| *** tojuvone has joined #openstack-fenix | 10:37 | |
| *** tojuvone has quit IRC | 10:48 | |
| *** tojuvone has joined #openstack-fenix | 10:48 | |
| *** tojuvone has quit IRC | 10:51 | |
| *** tojuvone has joined #openstack-fenix | 10:51 | |
| *** tojuvone has quit IRC | 10:58 | |
| *** tojuvone has joined #openstack-fenix | 10:59 | |
| *** tojuvone has quit IRC | 12:17 | |
| *** tojuvone has joined #openstack-fenix | 12:18 | |
| *** tojuvone has quit IRC | 12:55 | |
| *** tojuvone has joined #openstack-fenix | 12:55 | |
| *** tojuvone has quit IRC | 13:04 | |
| *** tojuvone has joined #openstack-fenix | 13:04 | |
| *** tojuvone has quit IRC | 13:45 | |
| *** tojuvone has joined #openstack-fenix | 13:46 | |
| *** tojuvone has quit IRC | 13:47 | |
| *** tojuvone has joined #openstack-fenix | 13:48 | |
| *** tojuvone has quit IRC | 14:09 | |
| *** tojuvone has joined #openstack-fenix | 14:09 | |
| *** tojuvone has quit IRC | 14:24 | |
| *** tojuvone has joined #openstack-fenix | 14:25 | |
| *** tojuvone has quit IRC | 14:31 | |
| *** tojuvone has joined #openstack-fenix | 14:32 | |
| *** tojuvone has quit IRC | 14:39 | |
| *** tojuvone has joined #openstack-fenix | 14:39 | |
| *** tojuvone has quit IRC | 14:43 | |
| *** tojuvone has joined #openstack-fenix | 14:44 | |
| *** tojuvone has quit IRC | 15:05 | |
| *** tojuvone has joined #openstack-fenix | 15:05 | |
| *** tojuvone has quit IRC | 15:16 | |
| *** tojuvone has joined #openstack-fenix | 15:17 | |
| *** tojuvone has quit IRC | 16:27 | |
| *** tojuvone has joined #openstack-fenix | 16:27 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!