05:01:46 #startmeeting Fenix 05:01:47 Meeting started Mon Apr 8 05:01:46 2019 UTC and is due to finish in 60 minutes. The chair is tojuvone. Information about MeetBot at http://wiki.debian.org/MeetBot. 05:01:48 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 05:01:51 The meeting name has been set to 'fenix' 05:02:30 #topic status 05:02:59 Greetings from ONS.. 05:03:10 ONS demo went very well 05:03:18 awesome 05:03:18 There was a lot of comments that we need Fenix and big companies to visit and see demo 05:03:37 Yes, people might be doing rolling upgrades, but nobody has our interaction with VNF 05:03:49 hyunsikyang hi, are you there? 05:03:49 and is needed 05:05:18 oh yes, hyunsikyang 05:05:30 yes 05:05:31 Hi 05:06:03 hyunsikyang, Hi, are you able to be in meeting now? My basd we started 1hour early 05:06:15 Yes 05:06:17 I can 05:06:44 great :) So was having status fro ONS, will continue.. 05:06:56 okie 05:07:00 Still have not composed more than this video 05:07:04 #link https://youtu.be/NsNBtLA-18I 05:07:16 So this is with sound the upgrade demo 05:07:33 +1 cool 05:07:37 From the ONS presentation experience, I see there is need of doing a video that explains Fenix 05:07:39 Great:) 05:07:48 Planning to do that soon. Using some new slides and maybe including the live demo at the end. 05:08:21 It will be very important for those from ETSI and new people and to make the case for the whole industry 05:08:44 The VNF interaction and standard way of doing it 05:09:23 Any question from the ONS, or shall I continue to further status? 05:09:50 I don't have for ONS 05:09:54 No. Please go ahead. 05:10:12 ok.. 05:10:18 Demo related changes to Fenix are under review: https://review.openstack.org/648613 05:10:25 However that is purely the Fenix part. There is a lot of more related the actual DevSTack upgrade 05:10:41 I would come to that in next topic 05:11:11 ok, will try to review it today. 05:11:27 So, I would still have one status item from me.. 05:11:31 dangtrinhnt, great :) 05:11:41 While I have been discussing with ETSI, there is one thing I realized in ONS 05:11:57 With VNF there should be grouping of all different VMs and some constraints to this group 05:12:02 Like how many VMs at least need to be alive always (this should be dynamically updated for current situation) 05:12:15 So, I realize we should not discuss to have this by any means a change to Nova 05:12:27 Otherwise we loose the ability of Fenix to work on any cloud 05:12:36 Yes, this possible to have in Nova, but then proxied through Fenix, so VNF have only single access point 05:13:21 This basically is extending the Affinity / AntiAffinity groups 05:13:46 And definition will be coming from ETSI specs 05:14:31 So this is my opinion to have in Fenix any comments? 05:15:51 I will come later to this anyhow as there will be more of ETSI work later in the meeting 05:16:08 so there is place to discuss later in PTG 05:16:17 Where is discuss channel? 05:16:40 what channel you use for this discussion? 05:17:07 Need to think more about this. Btw, tojuvone, my visa application has been rejected. 05:17:45 I need a more detail information for this. I understood briefly now. but I need more detail info. 05:17:48 hyunsikyang, I come back to that in last topic. Meanwhile I have had one meeting with ETSI people 05:18:26 dangtrinhnt, oh, that is too bad :( is that final or are can you still get it? 05:18:47 Oh... sad to hear that 05:18:52 It's final :( have to cancel everything. 05:18:55 what happen? 05:20:09 Not sure why, I have all the document, financial support, flight ticket, invitation letter etc. the woman at the embassy just told me that I'm not qualified. 05:20:12 I think that purpose of your visit is definitely clear.. 05:20:27 humm... 05:20:28 sure 05:20:56 Hi 05:20:59 Knew there is a problem with Chinese to US, but not for others 05:21:06 JangwonLee_, hi :) 05:21:20 He is my coworker :) 05:21:30 maybe because I'm a Vietnamese but living in Korea. 05:21:32 anyway 05:21:34 It is also now something looked with other open source communities to see where events could be kept 05:21:44 so nobody would have problems of going 05:22:07 anyway, still can working from remote :) 05:22:23 ok, let's get back to the topic. 05:22:35 Ok 05:22:35 dangtrinhnt, yes, we should seek to have a remote connection 05:22:51 ok, getting back.. 05:23:42 I actually completed my status and getting back to that ETSI.. stuff later in meeting 05:24:01 That's great. 05:24:12 dangtrinhnt, Do you have any status to tell? 05:24:18 Ok. I also intereted in ETSI discussion. watinig your update. 05:25:43 Nothing except for tomorrow I will kick off the Telemetry team for the Train cycle. Hopefully will find some directions for Ceilometer, AODH. If you guys wanna have some input for Fenix-AODH relationship, put it here #link https://etherpad.openstack.org/p/telemetry-train 05:26:35 dangtrinhnt, great :) 05:26:53 good :) 05:27:02 dangtrinhnht, 05:27:22 Surely we have the needed alarm config, but still I do not know it needs changes to those projects 05:27:29 and tojubone 05:27:32 can we make a blueprint for Fenix AODH? 05:27:43 becasue it is about if use Fenix 05:28:12 It is fisrt step for tacker integration? 05:28:15 ! 05:28:25 and there is also the possibility to get rid of the AODH as mandatory piece 05:28:35 ok 05:29:29 VNFM could also subscribe to Fenix directly and that might be better for generic thing on different clouds 05:29:44 It makes sense 05:30:08 and what is in Ceilometer / AODH now, already works for Fenix use case 05:30:15 you mean that VNFM get the alarm form Fenix directly? 05:30:19 As long as Ceilometer/AODH makes release :) 05:30:47 tojuvone, I'll try :) 05:30:47 hyunsikyang, Yes, it might be future improvement 05:30:55 Yes. 05:30:56 dangtrinhnt, +1 05:31:19 So, whatif, we use AODH for tacker integration, we try to start it first 05:31:41 I found that there are still lots of interest in Telemetry so I don't think it gonna be a problem. 05:31:45 hyunsikyang +1 05:31:59 Yes, we go with this for now. There is nothing yet to replace the AODH for any means 05:32:12 OK:) 05:32:13 Just that I see it more generic to be without it at some point 05:32:30 Yes of course. 05:32:59 so great dangtrinhnt is working on these project :) 05:33:02 It is better than specific... 05:33:52 ok, so I have no more comment on this. 05:34:01 ok, thanks 05:34:12 hyunsikyang, any status to report? 05:34:21 I have. 05:35:01 Fisrt item is AODH issue. but, we discussed it already.. 05:35:15 So we will make a plan for it first, 05:35:29 And the second one is blueprint. 05:35:46 I updated blue print, please check it again:) 05:36:09 hyunsikyang, I will look today :) 05:36:17 Thanks:) 05:37:31 ok, so we can continue that in gerrit then 05:37:47 +1 05:38:21 +1 05:38:36 moving on then I guess 05:38:55 JangwonLee_, Do you have anything? 05:39:21 Nope. ok. 05:39:51 if not I could say some on csatari behalf 05:40:35 sure, please. 05:40:51 So he has done some documentation for us and might continue some later 05:41:16 And he is helping me with ETSI 05:41:22 cool 05:41:49 and still... coming to that in last topic 05:42:05 ok 05:42:15 so, would more to next topic to get meeting in time.. 05:42:29 #topic Testing 05:42:47 Demo had many changes we need that needs a way to "upstream" 05:42:54 ONS Demo had new action plugins to upgrade DevStack 05:42:59 It changes Queens and rocky DevSTack scripts 05:43:05 It needed changes to DevStack target, node specific configs and scripts 05:43:14 It also had huge changes to Doctor testing scripts, to be able to run multitenant and on Contoller 05:43:23 I would like to share this, but I haven't figured out how. Maybe by building it to Fenix test? 05:43:35 Current problem is also that do we really want to have OpenStack upgrade in our context? 05:43:46 It takes currently one hour and proper implementation with more than one controller even more 05:43:51 I mean; Fenix can be utilized for any maintenance use case, but implementing those might be overkill? 05:44:01 But, OpenStack upgrade is still the thing for the community? 05:44:08 Should we have this one case as example or use just dummy action plugins in testing? 05:44:14 Dummy plugins would test Fenix itself as sanity, OpenStack upgrade is something else? 05:44:23 Thoughts? 05:45:23 Dummy also not bed to show the fenix works. but if we find a simple action to show the fenix work efficiently.. it is good. 05:45:58 Yes, maybe "dummy" as normal gerrit job... 05:46:02 I will think about it and tell it here. 05:46:17 but a way to show also the OpenStack upgrade 05:46:36 or to have the openstack upgrade too, as non voting? 05:47:01 Other projects still working on a neat way to upgrade so Fenix is a great solution I guess. 05:47:19 non-voting for now until we make it to work, I guess. 05:47:41 yes, but still I think I shoudl have all that hacked DevStack thingy in Fenix repo if test with it 05:47:48 For testing we can setup some special testing configuration for Fenix vs OpenStack 05:48:25 yes, I think there shoudl be 2 environments we can run with gerrit patches then 05:48:33 zuul support that as fas as I know. 05:48:38 voting one with dummy action plugins 05:48:58 and non voting that upgrades OpenStack 05:49:46 dangtrinhnt, great. I also do not know yet anything about the zuul, so it will be learning curve to make that automation 05:49:59 ok, I think there is a plan then :) 05:50:19 anythign else on this, or moving on? 05:50:26 I am done 05:50:59 please, move on. 05:51:01 #topic Summit and PTG 05:51:10 Open Infrastructure summit: Fenix - Rolling Maintenance and Upgrade forum session April 30th 05:51:16 Planning page: https://etherpad.openstack.org/p/DEN-fenix-forum-brainstorming 05:51:25 Have to work on this. Problem is mostly that have to tell about Fenix.. 05:51:29 but have time for project updates and discussion 05:51:58 Straight forward otherwise 05:52:17 +1 05:52:23 I will join this meeting:) 05:52:25 Maybe it shall help as I am working on those slides to have a video to explain Fenix 05:52:26 +1 05:52:43 So should be able to show it faster 05:53:10 good:) 05:53:17 ok, so PTG then.. and now comes the ETSI thing 05:53:37 Fenix PTG: May 4th 05:53:43 PTG planning page: https://etherpad.openstack.org/p/DEN2019-fenix-PTG 05:54:00 Have to work this also. Can prioritize some there, but it is hard to fore see do we have more developers or not 05:54:00 Hard to know what is possible in Train and how much and when get to ETSI specific changes 05:54:22 but.. 05:54:33 That is not the only Fenix PTG session 05:54:41 There should also be ETSI FEAT03 discussion with Fenix 05:55:00 This means there is 2 Fenix PTG sessions. Expecting 1h + 1h or so. 05:55:11 csatari will help in organizeing the ETSI discussion 05:55:19 +1 05:55:41 and in that session it will be discussed of those "VM gouping" that I was telling about 05:55:56 VM grouping 05:56:30 To Nova or Fenix or FEnix, but proxied to Nova 05:57:50 Among discussing all other details of FEAT03. Should Fenix be changed accordingly, or the specfications in ETSI 05:58:16 That will be a tons of fun :) 05:58:24 Exciting 05:58:24 Proxied to Nova means Nova still have to do the job? As fas as I understand, Nova has its own way of rolling upgrade. https://docs.openstack.org/nova/latest/user/upgrade.html 05:58:37 I 05:58:49 I'm more of changing Fenix than ETSI 05:59:23 I have read that document 05:59:38 And tried to follow it in Fenix action plugins 06:00:15 Upgrade needs Controllers first and I do it 06:00:19 then computes 06:00:40 and following that document is how Nove needs to be done 06:00:55 oh.. running out of time.. 06:01:09 fast.. 06:01:17 #topic AoB 06:01:23 Meeting time? 06:01:34 No meeting 6th May as just discussed in Denver 06:01:41 Updating the meeting info to indicate this: https://wiki.openstack.org/wiki/Fenix#Meetings 06:02:04 How about meeting time? 6 UTC or 7? 06:02:14 oh 6 or 5? 06:02:38 Today we had it 5, but do we keep the 6 06:02:49 5 would be better for me? 06:03:34 5 keeps it in same place for me after daylight saving changed. But can keep the current 6 too 06:03:49 ? 06:04:08 to me both is good 06:04:30 dangtrinhnt 06:04:47 5 06:04:49 :) 06:04:54 thanks 06:05:12 ok :) I will upfdate 5 to wiki :) 06:05:27 Anything else or do we close? 06:05:39 I'm good. Thanks. 06:06:08 closing the meeting then, thanks for joining :) 06:06:18 #endmeeting