14:02:18 #startmeeting Fedora IoT Working Group Meeting 14:02:18 Meeting started Wed Aug 14 14:02:18 2019 UTC. 14:02:18 This meeting is logged and archived in a public location. 14:02:18 The chair is pbrobinson. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:02:18 Useful Commands: #action #agreed #halp #info #idea #link #topic. 14:02:18 The meeting name has been set to 'fedora_iot_working_group_meeting' 14:02:19 #chair pwhalen pbrobinson bcotton 14:02:19 Current chairs: bcotton pbrobinson pwhalen 14:02:19 #topic roll call 14:02:29 .hello2 14:02:30 bcotton: bcotton 'Ben Cotton' 14:02:38 .hello2 14:02:39 lorbus: lorbus 'Christian Glombek' 14:02:41 * pwhalen here 14:02:57 * tdawson is here too. 14:03:14 * lorbus as well 14:03:38 haven't attended in a while, good to see y'all =) 14:04:08 you too, lorbus :) 14:04:19 #topic 1) ==== Working Group process and admin ==== 14:04:59 so I think the big things here are edition planning 14:05:14 #link https://docs.fedoraproject.org/en-US/iot/edition-promotion/ 14:05:22 I need to update this doc 14:05:34 I had a meeting at flock with CPE about some of the items 14:08:17 does any one have any queries 14:08:48 I see for the targets it sounds like aarch64 containers aren't being built. I thought they were. 14:09:27 I don't remember who I asked about that, but I distinctly remember asking someone if we had aarch64 containers 14:09:45 nope, there's base containers being pushed to docker, this is about the build pipeline, so build something like mosquitto as a container and submit it as an update via bodhi 14:10:07 tdawson: base containers isn't the same as a full application stack container 14:10:18 OK ... so maybe, whoever I asked was thinking that. Makes sense. 14:10:32 Either way, armv7 still needs it's containers too. 14:10:50 yes, it's on my list 14:11:04 I need to spend time to work out the issues with the base containers and fix that again 14:11:40 Other question. Do we have a list of what we want in thet "IoT Toolbox"? 14:12:26 mostly I want to put a bunch of the utils that we currently have in the base image in there 14:12:38 at least as a start 14:12:47 OK .... makes sense. 14:13:12 #topic 2) ==== Fedora 31 ==== 14:13:26 so F-31 has branched off rawhide this week 14:13:36 we don't yet have a main compose 14:14:04 but when the primary compose is working I'll kick off the F-31 and rawhide IoT processes and get them in place 14:14:41 #info the devel branch which has had F-30 on it will become F-31, if you've not rebased an earlier F-30 to stable branch you'll move to F-31 if you update 14:15:23 any one have anything for F-31 14:15:45 I've been playing around with Node-Red on F-31 IoT, it's now running but does need some work 14:16:35 Cool 14:18:22 no one with any thing for F-31 at all? pwhalen any news on CI/CD bits? 14:18:52 pbrobinson, working on adding iot to openqa tests 14:19:17 hopefully have some tests running this week is my goal 14:20:10 I've rebased the greenboot preset PR in fedora-release. Has anyone tested else it by enabling them manually? 14:20:10 #topic 3) ==== Fedora 30 ==== 14:20:15 #undo 14:20:15 Removing item from minutes: 14:20:35 lorbus: not really, but I have people wanting it so it will be RSN 14:20:48 it's been on my list but I just need more time :) 14:21:10 ack. I have a rather tight schedule at the moment, but definitely let me know if I can do anything to help! 14:23:50 #topic 3) ==== Fedora 30 ==== 14:24:15 so the F-30 stuff seems fine ATM, we're on 5.2 and it's published to the stable branch 14:24:36 the people I've been testing it with haven't reported any issues 14:26:35 no one done anything? 14:28:00 Other than basic running, nothing. 14:28:43 I've got some containers running on my f30 iot host, no issues so far 14:29:02 #topic 4) ==== Flock Budapest 2019 ==== 14:29:18 A quick report on Flock in the context of IoT 14:29:26 there was a number of good talks 14:29:41 the two I was involved in were good, and had some great feedback 14:29:54 * pwhalen is hoping they show up online soon 14:29:54 the first will be published online with video and slides 14:30:32 the second one we didn't have authorisation to publish, I co-presented with Chad from ExxonMobil the work they're doing with me on Fedora IoT 14:31:03 so that's pretty cool that they're now publicly talking about how they're using it and why 14:31:33 a lot of people were quite shocked at what we were doing, the issues involved and how we were solving them 14:31:54 I assume it will be an internal product for them? Not something they will be selling. 14:32:16 correct, they're using it directly in a number of use cases 14:32:29 very nice 14:33:33 and puiterwijk gave a talk on the ongoing work we're going to be doing in Fedora IoT as part of his new IoT Security role 14:33:40 so welcome to the IoT team puiterwijk 14:34:07 and that will be published too 14:34:25 Welcome, puiterwijk :-) 14:34:26 awesome! 14:34:34 Thanks :) 14:35:35 o/ Patrick :) 14:37:00 puiterwijk: do you, or anyone else that was there, jsmith etc, have any comments on flock? 14:39:05 Not that I can think of 14:39:20 #topic 5) ==== Open Floor ==== 14:40:30 anyone with general business? 14:40:40 I have a question for Open Floor. 14:40:49 (Sorry, a little slow here -- I talked to several people at Flock to get their reactions, and I think everyone was pleasantly pleased with progress, and surprised with the traction Fedora IoT is getting inside industry.) 14:41:16 #undo 14:41:16 Removing item from minutes: 14:41:40 thanks jsmith, that was the general feeling I got as well 14:42:10 #topic 5) ==== Open Floor ==== 14:42:13 lorbus: go 14:42:34 Are there any plans to move to the Fedora CoreOS build infra, and away form Anacaonda to Ignition? I heard some chatter recently, but didn't hear of any conclusion 14:42:49 (I mean in the veery long term) 14:42:58 lorbus: there's a bunch of stuff ignition can't currently do 14:43:26 I would like to move towards it for somethings but I find the build process for CoreOS insane and unsuitable for a lot of stuff 14:43:51 interesting. is there a doc that summarizes those shortcomings? 14:44:17 nope, because that would require me to spend days analysing and writing up and I don't have the time to do that 14:44:35 because as soon as it's documented people will throw stones at me 14:44:54 and frankly I have enough problems as it is 14:44:55 fair enough :) 14:45:01 Just want people to know that gateway 0.9 is out. It will be a couple of weeks before I'm able to get it updated. Also notejs-nanomsg (needed by mozilla-iot-gateway) isn't building on F31. Hopefully I can get that fixed this week. 14:45:19 tdawson: thanks 14:46:14 pbrobinson: From the other conversation, it looks like you are trying to package node-red ... it will be a couple of weeks before I have the time to help, but if you are still having issues packaging it then, I can give you a hand. 14:46:29 lorbus: there's a bunch of stuff interesting to me around ignition for provisioning but it currently doesn't do things like that we consider blocking so it's going to take time to analyse that and write up RFEs and get it so that the leads even care 14:46:41 pbrobinson++ nonetheless thanks for the info! glad to hear that hasnt dropped off the radar entirely, and I'm totally fine with it being something that is not planned in the short term 14:46:41 lorbus: Karma for pbrobinson changed to 5 (for the current release cycle): https://badges.fedoraproject.org/tags/cookie/any 14:46:56 ack 14:47:17 the provisioning part is why I'm interested in having it, too 14:47:20 lorbus: ignition would be nice because it would cut our image size by a large chunk 14:47:42 kill the python 14:47:49 and the ability to add things like cloud-init style support would be useful for use on AWS etc 14:48:04 lorbus: we won't be killing python because we need ansible 14:48:28 ah well 14:48:50 and speaking with people at Flock on the CoreOS side of things the "just tear down the node and redeploy" is fine when you're in a cluster in a datacentre.... 14:49:48 but is frankly a ridiculous stance in all other circumstances, basically fore CoreOS aiming to be a OS purely for running k8s it's fine, but for everyone else there's a need of python 14:50:08 yup. Rollback is an important feature, especially in IoT where you cant push a new image for each action over the wire 14:50:32 well especially in cases where there's a total bandwidth of 64K 14:51:14 same goes for their "update model" which is weird that they have a means to update the OS, when they're stance for changes is to just redeploy, why not just redeploy for updates too 14:52:32 I can answer that one, because we hit it on the OpenShift team. Because some customers freak out if they can't do updates. 14:54:24 the same customers will freak out when they can't deploy updates/config changes with ansible but ¯\_(ツ)_/¯ not my problem, out of scope for this meeting 14:54:45 anyone with anything else to add? 14:55:25 there is a difference between the update models of RHCOS (container-embedded ostree) and FCOS (update from Ostree repo, as we know it) tho 14:55:25 Nothing from me 14:55:57 I'll bring this up again a year from now or so :P 14:58:38 lorbus: drink all the cool aid you want ;-) 14:58:43 thank you all 14:58:43 #endmeeting