*** hyunsikyang has joined #openstack-fenix | 02:02 | |
tojuvone | Good morning, meeting in 10min | 04:50 |
---|---|---|
dangtrinhnt | tojuvone, hi | 04:51 |
hyunsikyang | hi | 04:53 |
tojuvone | dangtrinhnt, hyunsikyang hi :) | 04:53 |
tojuvone | #startmeeting Fenix | 05:00 |
openstack | Meeting started Mon May 20 05:00:10 2019 UTC and is due to finish in 60 minutes. The chair is tojuvone. Information about MeetBot at http://wiki.debian.org/MeetBot. | 05:00 |
openstack | Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. | 05:00 |
*** openstack changes topic to " (Meeting topic: Fenix)" | 05:00 | |
openstack | The meeting name has been set to 'fenix' | 05:00 |
dangtrinhnt | o/ | 05:00 |
tojuvone | Wecome to Fenix meeting, time to start | 05:00 |
tojuvone | So, let's have the first topic | 05:01 |
tojuvone | #topic status | 05:01 |
*** openstack changes topic to "status (Meeting topic: Fenix)" | 05:01 | |
tojuvone | Some bits from OIS + PTG... | 05:02 |
tojuvone | OIS summit and PTG went well, once more links to etherpads | 05:02 |
tojuvone | https://etherpad.openstack.org/p/DEN-fenix-forum-brainstorming | 05:02 |
tojuvone | https://etherpad.openstack.org/p/DEN2019-fenix-ETSI-NFV-PTG | 05:02 |
tojuvone | https://etherpad.openstack.org/p/DEN2019-fenix-PTG | 05:02 |
tojuvone | There was good amount of participants in forum and also good input in PTG | 05:02 |
tojuvone | 22 people in Forum was nice for a new project | 05:03 |
dangtrinhnt | wow | 05:03 |
hyunsikyang | Great | 05:03 |
tojuvone | and in PTG some 5 - 10.. wel there wass 2 sessions anyhow | 05:04 |
tojuvone | There was some discussion about the ETSI changes | 05:05 |
tojuvone | But did not have participants from ETSI side really | 05:05 |
tojuvone | Anyhow good input on the document proposal I had | 05:05 |
tojuvone | Like the interface to be used between Fenix and VNFM | 05:06 |
tojuvone | Otherwise the details are still not ready in ETSI too | 05:07 |
tojuvone | But the interface is the thing to be considered. So still with the AODH | 05:07 |
dangtrinhnt | I'm looking into it. Truth is AODH is only for OpenStack. | 05:08 |
hyunsikyang | Yes. I think the first integration of VNFM, I only think about using AODH and tacker API | 05:08 |
hyunsikyang | However, after that, we can make it independantly. | 05:08 |
hyunsikyang | with this, we also consider ETSI sol interface too:) | 05:09 |
tojuvone | There is was also this etcd v2 "watch" API that one can look at | 05:09 |
tojuvone | and still there is wither someting should be pushed to Nova | 05:10 |
tojuvone | bad thing is that it will then be Nova and not for other cloud | 05:10 |
dangtrinhnt | You meant this #link https://coreos.com/etcd/docs/latest/learning/api.html#watch-api | 05:11 |
tojuvone | Those are like "instance group" details during mainteannce | 05:11 |
tojuvone | dangtrinhnt, yes.. I looked fast there, but haven't given a thought on this | 05:12 |
tojuvone | The thing was that infra should nto call something external (VNFM) directly | 05:13 |
tojuvone | Anyhow just like if have opinions, this is what was mentioned | 05:14 |
tojuvone | Anyhow I would have reasons to have something else as priority first | 05:14 |
hyunsikyang | ok.. | 05:15 |
tojuvone | as becasue summit input... | 05:15 |
tojuvone | Want to hear? :) | 05:15 |
dangtrinhnt | So is it that you worrying about Fenix too much depends on OpenStack? | 05:15 |
tojuvone | Well, Kubecon is 16000 people | 05:16 |
dangtrinhnt | holy... | 05:16 |
hyunsikyang | kkkkkkkkk | 05:16 |
tojuvone | that is some number to play against | 05:16 |
tojuvone | Yes, we need to have the OpenStack, but that is so coming next | 05:17 |
hyunsikyang | right. | 05:18 |
tojuvone | In other words, whiel I have not had time for k8s, it will come and fast | 05:18 |
tojuvone | But maybe moving on from this ETSI thing for now? | 05:19 |
dangtrinhnt | True. I had a plan to add support for K8S in Searchlight this cycle. | 05:19 |
dangtrinhnt | +1 tojuvone | 05:19 |
tojuvone | And what would lead to our priority... | 05:20 |
tojuvone | As from PTG input... | 05:20 |
tojuvone | Two big users told to be interested in Fenix. This means we need to focus on maturity | 05:20 |
tojuvone | Like if I recall 147 regions were mentioned | 05:21 |
tojuvone | I would say as of this: Serving potential users is the first priority | 05:22 |
dangtrinhnt | Okie, I looked at the pad. So I think next thing we need to do is to define what maturity means for Fenix. | 05:22 |
tojuvone | yes | 05:22 |
dangtrinhnt | How can we know if Fenix is matured. Some measurable metrics like features, test cases, use cases, number of contributors, etc. | 05:23 |
tojuvone | For me it would be security (interfaces) and ability to take Fenix into use in general | 05:23 |
hyunsikyang | IMO, usecase is important | 05:23 |
tojuvone | Maturity as being able to utilize it. Having testing... | 05:24 |
hyunsikyang | so, I think that we focus on integration of openstack and we should show the result accroding to fenix architecture. | 05:24 |
tojuvone | yes, both user were surely looking into OpenStack maintenance scenario | 05:25 |
tojuvone | This bigger cloud case was to have continuous maintenance as of size of cloud | 05:26 |
tojuvone | So what is "generic" in Fenix should be in shape | 05:26 |
hyunsikyang | we show demo and open the source code and many people can test it. After mature of each function, we make it independently | 05:26 |
hyunsikyang | for generic. | 05:27 |
tojuvone | Whiel users might end up with own plugins | 05:27 |
tojuvone | Our biggest problem is to build testing | 05:27 |
tojuvone | Effort to complete that in every level is huge | 05:28 |
dangtrinhnt | true. | 05:28 |
tojuvone | So I have tried to put priority in storybord and wiki | 05:29 |
dangtrinhnt | okie, that's great. | 05:29 |
hyunsikyang | ok:) | 05:29 |
tojuvone | that would aim to have things ready for users | 05:29 |
dangtrinhnt | Thanks tojuvone. | 05:30 |
tojuvone | storyborad had some tag on priority work | 05:30 |
tojuvone | <level>-priority | 05:31 |
tojuvone | Latest work on this is live migration and almost ready devstack plugin | 05:31 |
tojuvone | Next would be documenting how to install Fenix and how to configure AODH for VNFM interaction | 05:31 |
tojuvone | but I have some other thing going on as well.. | 05:32 |
tojuvone | might slow me down, but again showing Fenix to the world | 05:32 |
dangtrinhnt | tojuvone, let put it there and let me see if can do something. | 05:33 |
tojuvone | dangtrinhnt, +1 | 05:33 |
hyunsikyang | yes. please share it:) | 05:33 |
tojuvone | So in OPNFV, looking to have a closed-loop demo for something "new" | 05:34 |
tojuvone | Work has started for self-healing + maintenance demo | 05:34 |
tojuvone | Maybe: Prometheus + Vitrage + Mistral and finally Fenix | 05:35 |
tojuvone | aim might be ONS Europe | 05:35 |
dangtrinhnt | great | 05:35 |
hyunsikyang | good | 05:35 |
tojuvone | This kind of closes everything OPNFV Doctor is aiming to do | 05:36 |
tojuvone | that is "fault management and maintenance" | 05:36 |
tojuvone | First you heal instance that is affected by memory fault | 05:36 |
tojuvone | isolate that memory, do nto allow new instances to host | 05:37 |
tojuvone | other instances still stable on host | 05:37 |
tojuvone | but Fenix needs to be called so one can maintain the host then | 05:37 |
tojuvone | and finally all instances running on healthy hosts | 05:37 |
dangtrinhnt | +1 | 05:38 |
hyunsikyang | what is the usage(function) of vitrage in this demo? | 05:39 |
tojuvone | I will keep you up-to-date on this | 05:39 |
hyunsikyang | just for GUI? | 05:39 |
tojuvone | Vitrage is root casue analysis | 05:39 |
tojuvone | so it analyses 5 memory faults from monitoring... | 05:40 |
tojuvone | raises alarm in its graph | 05:40 |
tojuvone | and Vitrage template will trigger further action on this | 05:41 |
hyunsikyang | using mistarl. I got it. | 05:41 |
tojuvone | Also Vitrage graph should show Fenix state for a host | 05:41 |
tojuvone | Mistral might have the workflow for self-healing | 05:41 |
tojuvone | While I might rather do the switchover for VM withing sample VNFM code | 05:42 |
tojuvone | as this should be using some Doctor test code | 05:42 |
tojuvone | This is part of Intel started OPNFV closed loop WG work | 05:43 |
tojuvone | ok, I do not think I have anything further | 05:44 |
tojuvone | Do you have something? | 05:44 |
dangtrinhnt | The telemetry is designing its new future. If you have any input, please feel free #link https://etherpad.openstack.org/p/telemetry-train-roadmap | 05:45 |
tojuvone | Great, I also saw there was some AODH demo.. need to look that | 05:47 |
tojuvone | autohealing | 05:47 |
dangtrinhnt | Some suggest auto-scaling and self-healing that may need Fenix :) | 05:47 |
dangtrinhnt | This #link https://www.youtube.com/watch?v=dXsGnbr7DfM&feature=youtu.be | 05:48 |
tojuvone | yes, you could do "self-healing" with Fenix too | 05:48 |
tojuvone | I just do not see it in Telco thing where you react fast | 05:49 |
dangtrinhnt | lxkong, one of the core made that. We're developing some use cases of Ceilometer and AODH in Train-1 milestone | 05:49 |
tojuvone | so more like maintenance | 05:49 |
tojuvone | :) | 05:49 |
dangtrinhnt | :) | 05:49 |
tojuvone | I will definitely watch that | 05:50 |
dangtrinhnt | But I think it more like application self-healing than infras. | 05:50 |
tojuvone | yes, self-healing is mostly that | 05:51 |
tojuvone | if you are not in a hurry, you can utilize Fenix | 05:51 |
tojuvone | Otherwise you need something faster | 05:51 |
dangtrinhnt | Absolutely, I will put it on the next meeting agenda of Telemetry. | 05:51 |
tojuvone | Normally you need ready policy fro infra fault to trigger self-healing | 05:52 |
tojuvone | in split second | 05:52 |
tojuvone | but if it is not that crucial, it can be done with mainteannce session, like fenix | 05:53 |
dangtrinhnt | +1 | 05:53 |
dangtrinhnt | Fenix is on our meeting agenda this Thursday :) #link https://etherpad.openstack.org/p/telemetry-meeting-agenda | 05:53 |
hyunsikyang | Tacker also can use it later. | 05:54 |
tojuvone | great, let's see if I can join the meeting | 05:54 |
hyunsikyang | good | 05:54 |
tojuvone | Yes, you have the Tacker BP ongoing :) | 05:54 |
dangtrinhnt | Okie, that's all from me. | 05:55 |
tojuvone | ok, anything else? I will need to jump to doctor meeting soon | 05:55 |
hyunsikyang | nothing from me. | 05:56 |
tojuvone | ok, thanks for the meeting | 05:56 |
tojuvone | And if want to work on sometinhg, just ask | 05:56 |
dangtrinhnt | Thanks, tojuvone. | 05:56 |
hyunsikyang | thanks all | 05:56 |
tojuvone | #endmeeting | 05:56 |
*** openstack changes topic to "Welcome to Fenix. Next meeting: 20th May 5 AM UTC. Add topics to: https://wiki.openstack.org/wiki/Fenix#Meetings" | 05:56 | |
openstack | Meeting ended Mon May 20 05:56:58 2019 UTC. Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4) | 05:57 |
openstack | Minutes: http://eavesdrop.openstack.org/meetings/fenix/2019/fenix.2019-05-20-05.00.html | 05:57 |
openstack | Minutes (text): http://eavesdrop.openstack.org/meetings/fenix/2019/fenix.2019-05-20-05.00.txt | 05:57 |
openstack | Log: http://eavesdrop.openstack.org/meetings/fenix/2019/fenix.2019-05-20-05.00.log.html | 05:57 |
openstackgerrit | Tomi Juvonen proposed x/fenix master: Add DevStack plugin https://review.opendev.org/659793 | 07:38 |
openstackgerrit | Tomi Juvonen proposed x/fenix master: Add DevStack plugin https://review.opendev.org/659793 | 11:48 |
tojuvone | dtruong, I watched the summit presentation: https://youtu.be/e9Mj7k-BoGU | 12:27 |
tojuvone | That was great :) | 12:27 |
tojuvone | So I have now had some input from other huge cloud that could look into Fenix | 12:31 |
tojuvone | so looking to constantly running Fenix maintaining some hosts in some part of the cloud. | 12:32 |
tojuvone | Also I bet have to be very careful with the DB. Anyhow currently only keeping there instances under maintenance session | 12:33 |
tojuvone | Not like "whole inventory". Anyhow not having the whole inventory is that, Fenix do not hold the history of what host has undergone which maintenance over time | 12:36 |
openstackgerrit | Merged x/fenix master: Add DevStack plugin https://review.opendev.org/659793 | 12:54 |
*** tojuvone has quit IRC | 13:35 | |
*** tojuvone has joined #openstack-fenix | 13:36 | |
dtruong | tojuvone Glad you liked the presentation :) | 15:58 |
dtruong | For our use case of hypervisor maintenance, we are thinking of just using different host aggregates that contain the maintenance information (i.e OS version, firmware version, etc) as metadata. | 16:00 |
dtruong | That way we don't have to keep track of which hosts have been updated or not in a separate DB. | 16:02 |
*** tojuvone has quit IRC | 19:58 | |
*** tojuvone has joined #openstack-fenix | 19:59 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!