15:00:44 #startmeeting openstack-helm 15:00:44 Meeting started Tue Apr 10 15:00:44 2018 UTC and is due to finish in 60 minutes. The chair is mattmceuen. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:00:45 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:00:48 The meeting name has been set to 'openstack_helm' 15:00:58 #topic Rollcall 15:01:04 Hey roman_g! 15:01:07 o/ 15:01:09 o/ 15:01:15 o 15:01:16 o/ 15:01:28 o/ 15:01:30 o/ 15:01:32 https://etherpad.openstack.org/p/openstack-helm-meeting-2018-04-10 15:02:10 Please add in all the burning topics, questions, etc you would like to discuss today 15:03:24 I'm listening in. Mostly. 15:03:40 #topic Multiline LMA functionality (srwilkers and jayahn) 15:03:59 srwilkers and jayahn -- anything left to discuss on this one or did you level set offline? 15:05:00 ive been working towards something that works here 15:05:40 initially, i thought using something like the journald logging driver would close the gap here. after some digging though, using anything but the json-file logging driver results in a loss of the ability to query for logs via kubectl 15:05:43 which is a regression 15:06:23 this makes me sad panda 15:06:30 portdirect: tres sad 15:07:10 there are open issues in kubernetes that im watching for any updates, but seems the ability to use journald along with kubectl was removed when CRI was introduced 15:07:10 what's the downside of json-file approach? 15:07:45 downside of the json-file approach is that we need to define regex's for each log format we may encounter with osh/osh-infra 15:08:01 fluentbit will need a regex to identify the first line in a multiline entry (like stacktraces, etc) 15:08:14 just requires some additional work on our end 15:08:50 gotcha 15:08:57 there was some work done a long time ago to try and address this (dulek started it forever ago) that we might be able to leverage as a base if necessary 15:09:40 but really, that's all ive got to say about it 15:09:50 thanks srwilkers 15:10:11 can we stream logs to the network and capture/process them on separate logging host? kubectl is kinda not supposed to be used for digging into logs. Only for fast lookups. 15:10:48 I'm not very competent in the topic, you may ignore question )) 15:10:57 roman_g, that's what fluenbit is doint actually 15:11:03 roman_g: thats the idea of having a centralized logging tool like fluentbit. 15:11:25 srwilkers, does the journald driver account for multiline though ? 15:11:50 yes, journald works well with multiline logs 15:11:51 cristicalin: it seemed to handle it better in the initial work i was doing 15:12:00 as the stdout/err from containers still gets broken into multiple lines 15:12:27 so docker would still see multiple lines and not know they are from the same message 15:12:38 or you use openstack loggins to send to journald ? 15:13:14 on our end we force openstack log formatting and have looked into using a json formatter 15:13:45 works nicely for openstack originated logs but goes into a recursive crash loop for things like sqlalchemy 15:14:26 yikes 15:15:24 here's what we use now, ignore the consul-template bit http://paste.openstack.org/show/718845/ 15:16:54 Just curious cristicalin, what drove you guys to the journald approach? 15:17:16 mattmceuen, we don't use journald, just experimented with it 15:17:22 we use json-file 15:17:58 ahh gotcha 15:19:13 cristicalin: neat. will take a gander later today 15:19:22 srwilkers do we have a good list of all the multiline formats we'll need to support regexes for? 15:19:33 and/or just address them as we run accross them? 15:20:25 mattmceuen: not at the moment. the openstack services would be the ones to tackle first, then can address the other ones bit by bit 15:20:36 Cool - sounds good to me, thanks 15:20:44 #topic Storyboard for 1.0 integration items 15:21:15 As you guys know, the Foundation is working to move most projects to use Storyboard instead of Launchpad this cycle 15:21:49 At the PTG we decided we would go with the flow here, rather than being a trailblazer or a laggard 15:22:15 That still means we should start planning for it and find the right time to migrate 15:22:51 We have our 1.0 requirements spec in progress: https://review.openstack.org/#/c/556325/ 15:23:14 That will result in a lot of granular bits that need to be coordinated well and worked quickly across our entire team 15:23:20 So we can get to 1.0 :D 15:23:52 I'm wondering, would it make sense to move to storyboard before we create all those work items, so we don't create them and then immediately migrate them from one system to another? 15:24:05 How is that related to storyboard? Or storyboard will be used to get to 1.0? 15:24:21 storyboard in this context is just a place to put the work - no direct tie in 15:24:25 ok 15:26:18 opinions welcome :) If you're not familiar with storyboard yet, it's an openstack scope management tool created within openstack with openstack-like workflows and collaboration in mind 15:26:19 https://wiki.openstack.org/wiki/StoryBoard 15:26:38 Be interesting to see which projects have converted over. I know I set it up for kolla, but we didn't go to town on it - mainly because bug-tracking seemed way off. 15:27:12 octavia has added a few things on their storyboard but not much 15:27:17 https://storyboard.openstack.org/#!/project_group/list 15:27:26 ^ projects that have migrated so far 15:28:16 Cool. 15:29:10 I think one thing that would be valuable would be for one of our team to do something of a deep dive into storyboard, see how other projects use it, and help guide OSH into using it well 15:29:36 Anyone interested in taking that on? 15:29:49 Has it been enabled for osh yet? 15:30:33 mattmceuen: be happy to take that on, seems like busy-work that could get me more involved in osh community 15:30:37 Somehow an osh project was already created, it's a mystery... I think we'll just recreate it upon migration I think 15:31:05 lol rwellum - that would be much appreciated and I think it'll actually be extremely valuable 15:31:12 np 15:31:35 I've gotten feedback from multiple folks that OSH would really benefit from a better-defined community roadmap 15:31:49 And I think migrating to storyboard would be something of a fresh start 15:32:03 +1 15:32:21 Although you've just described all of OpenStack :) 15:32:47 Well in that case, maybe we can define some good conventions that others can benefit from ;-) 15:33:01 I have some things I will share with ya offline too to give more background 15:33:14 Talking about roadmap - I did add a relevant topic to the etherpad. 15:33:22 Sure sounds good - I need to talk to you anyway :) 15:33:50 nice -- let's segueway right into it, I just have one short thing for later 15:33:53 #topic Support for other releases? Pike, Queens etc 15:34:04 we do already 15:34:23 Nancy created storyboard for OSH 15:34:26 but still need to demonstrate upgrade in the gate i think before moving this into the repo 15:34:36 as agreed at the ptg 15:34:42 'the sugarcubes' 15:34:51 'the sugarcubes' 15:34:51 So I am demoing osh to my company tomorrow - you're talking 60+ people, one of the things they are going to ask is releases, and upgrades. 15:35:07 So that's good news portdirect 15:35:26 I've been running master again recently... 15:35:45 https://en.wikipedia.org/wiki/The_Sugarcubes ?? 15:36:02 almost as cool 15:36:20 I'm working on writing up the versioning spec for OSH 15:36:34 roman_g: cool, glad we know who did it - I'll look into it more. 15:36:49 portdirect: great. Let me know if you need a guinea pig? 15:36:52 and the simplest way to hash it out it was using sugar cubes on a desk 15:37:03 I'm expecting diagrams of sugarcubes in the spec portdirect 15:37:11 ascii 15:37:16 there was also a teaspoon 15:37:31 (to represent upgrade paths) 15:38:07 A nice cup of tea to represent falling into an upgrade whirlpool 15:38:18 no! 15:38:26 Our position overall rwellum is that we want to forge our way to the current release quickly, with gated upgrade support, and non-voting gates for master 15:38:30 openstack helm makes for nice calming tea 15:38:49 that may be drunk with the pinky out 15:38:57 :) 15:38:57 Feel free to share that with your crew and good luck with the presentation :) 15:39:04 ++ 15:39:05 mattmceuen:+1 15:39:33 Last thing on the agenda: 15:39:39 #topic IRC - Slack integration revisited 15:40:06 So a few weeks ago we lamented the fact that Slack was removing its gateway for IRC (etc) integration 15:40:21 Believeing our Slack-IRC integration would be SOL 15:40:49 I chatted with cw since then, and he says that the tooling we use for Slack-IRC integration should actually not be impacted by this, that it doesn't use that particular gateway 15:41:00 So that's excellent news :) 15:41:57 #topic Reviews Needed 15:44:20 Anyone have any PS that they'd like to share to get some eyeballs on them? 15:45:03 not at the moment 15:45:10 active/passive maria ? 15:45:39 actually 15:46:04 wouldnt mind some eyes on: https://review.openstack.org/#/c/498929/ 15:47:22 active/passive maria - was chatting w/ portdirect on that and he may have a thought on how to simplify the approach and get it done fast. Can we plan on that for a topic of discussion next meeting cristicalin? 15:47:44 yep, will try to make it 15:48:19 awesome - thanks man 15:48:28 I'll ping portdirect later during the week as we are also looking into the problem 15:49:03 perfect 15:49:15 #topic Roundtable 15:49:26 open discussion, anything else on your minds? 15:51:08 Alright guys - good meeting, and have a good week! See you in #openstack-helm 15:51:11 #endmeeting