18:00:07 <smooge> #startmeeting Infrastructure (2017-09-07) 18:00:08 <zodbot> Meeting started Thu Sep 7 18:00:07 2017 UTC. The chair is smooge. Information about MeetBot at http://wiki.debian.org/MeetBot. 18:00:08 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic. 18:00:08 <zodbot> The meeting name has been set to 'infrastructure_(2017-09-07)' 18:00:08 <smooge> #meetingname infrastructure 18:00:08 <zodbot> The meeting name has been set to 'infrastructure' 18:00:08 <smooge> #topic aloha 18:00:08 <smooge> #chair smooge relrod nirik abadger1999 dgilmore threebean pingou puiterwijk pbrobinson 18:00:08 <zodbot> Current chairs: abadger1999 dgilmore nirik pbrobinson pingou puiterwijk relrod smooge threebean 18:00:12 <smooge> Welcome everyone 18:00:17 * relrod waves 18:00:24 <nirik> morning 18:00:32 <laura_> hello 18:00:39 <nauseousSartre> hi 18:00:55 <saif_> hi 18:01:12 * threebean waves 18:01:13 * cverna waves 18:01:30 <smooge> topic New folks introductions 18:01:33 <marc84> hi everyone 18:02:03 <smooge> #info This is a place where people who are interested in Fedora Infrastructure can introduce themselves 18:03:34 <nauseousSartre> I am interested in contributing to Fedora Infrastructure. I program mainly in Python and it seems to me that I could be useful at Bodhi development 18:04:02 <smooge> Hello nauseousSartre 18:04:03 <laura_> hello everyone my name is Laura and I am very interested in the Fedora project as I want to learn more about system administration 18:04:09 <saif_> hi , my name is saif , i have some experience in linux and would love to join the infrastrucure group 18:04:14 <smooge> hllo laura_ and saif_ 18:04:22 <laura_> hello smooge 18:04:30 <nauseousSartre> Hi smooge 18:04:36 <marc84> welcome nauseousSartre , laura and saif_ 18:04:47 * nirik waves welcome to all the new folks. 18:04:48 <laura_> thanks marc84 18:04:58 <laura_> and thanks nirik 18:05:03 <saif_> #waves 18:05:18 <smooge> if you haven't already please send an email to the infrastructure@lists.fedoraproject.org list with information about your interests in systems admin or other topics 18:05:30 <smooge> we can add you to the apprentice group afterwords 18:05:41 <smooge> any other new people? 18:06:18 <laura_> will do smooge 18:06:46 <smooge> #topic announcements and information 18:06:46 <smooge> #info PHX2 Colo Trip, Dec 4th - 9th 18:06:46 <smooge> #info beta freeze started 2017-09-05 - everyone 18:06:46 <smooge> #info Moving .stg. ips ongoing 18:06:47 <smooge> #info DNS breakage happened earlier this week 18:06:48 <smooge> #info NFS downgrade from 4.1 to 4.0 complete 18:06:49 <smooge> #info NFS slowdowns not seen, but weird high IO has returned 18:06:53 <smooge> #info new pagure version deployed - pingou 18:06:55 <smooge> #info new bodhi in staging, please test - bowlofeggs 18:07:50 <smooge> Any other announcements? This week has been mostly catching up after FLOCK 18:08:04 <nirik> and stomping out fires. ;) 18:08:30 <smooge> we are currently in Beta freeze so changes to infrastructure are to be kept to a minimum and need to follow the Freeze Break process 18:08:52 <smooge> I don't know how long this freeze will be.. 18:09:37 <nirik> 19th or 26th or later. :) 18:10:36 <smooge> ok next up 18:10:49 <smooge> #topic flock recaps - everyone 18:11:08 <smooge> For those who got to make it to Flock it is now time to go over anything that needs to be reported for those who didn't 18:11:19 <smooge> Kevin you want to start it off? 18:11:32 <nirik> I plan to write up a list post on my workshop, but I can outline it here... 18:12:33 <nirik> basically I hope we can move more things to openshift down the road, also we should explore using some cloud providers for things that make sense there, also we should work together more with other open infrastructures like CentOS 18:13:01 * bowlofeggs lurks in the irc shadows 18:13:15 <nirik> There were a bunch of suggestions in the workshop to about specific things... like more metrics gathering, revist monitoring again, and so on. 18:13:44 * jcline eyes Sentry 18:13:48 <nirik> I'll have those things in my writeup. 18:14:29 <nirik> I also talked with CentOS folks and hopefully we can migrate our jenkins usage to centos-ci... 18:14:45 <nirik> they have more cycles to maintain jenkins than we do and have a better setup. 18:15:01 <pingou> big +1 18:15:14 <marc84> +1 18:15:21 * nirik tries to think of anything else mentioned... 18:16:03 <nirik> I'm sure there was a bunch more, but I don't have my notes handy... 18:17:04 <nirik> oh, one post flock note: 18:17:18 <bowlofeggs> nauseousSartre: reach out to me after the meeting in #fedora-apps and let's talk about python and infra :) 18:17:34 <nirik> it would be great if we could get some folks working on packages again. With pkgdb going away more people are using it and it needs a lot of love 18:17:42 <bowlofeggs> i've been using the centos jenkies and it does work pretty well 18:18:12 <nirik> bowlofeggs: there was thought of making it a supported CI in pagure too... ie, you could just fill out things there and that would enable it. 18:18:17 <bowlofeggs> nirik: perhaps that could be something nauseousSartre could help with too 18:18:19 <smooge> who owns(owned? owend?) packages 18:18:27 <bowlofeggs> nirik: that would be sweet 18:18:34 <nauseousSartre> bowlofeggs will do 18:18:43 <bowlofeggs> smooge: i always assumed it was pingou, so i will throw him under the bus! 18:18:46 <nirik> smooge: empty set I think. 18:19:00 <nirik> I think it was luke and ralph... 18:19:02 <bowlofeggs> yeah might be unmaintained 18:19:43 <nirik> anyhow, it: still runs on rhel6, still tries to query pkgdb for new packages and active branches and so forth, doesn't update descriptions or a bunch of things... 18:19:49 * cverna had a look at packages and seen some turbogears stuff :P 18:19:59 <nirik> yeah, it's TG1 I think 18:20:00 <bowlofeggs> oh my 18:20:04 <smooge> oof sounds like a rewrite 18:20:05 <bowlofeggs> sounds like a lot of work potentially 18:20:17 <nauseousSartre> could somebody explain what packages exactly was nirik referring to? 18:20:33 <nirik> https://apps.fedoraproject.org/packages/ 18:20:37 <bowlofeggs> this is packages: https://apps.fedoraproject.org/packages 18:20:45 <nirik> https://github.com/fedora-infra/fedora-packages 18:20:52 <nirik> it also uses moksha 18:21:00 <cverna> I think port to rhel7 is not to far of I had some note in the gobby 18:21:01 <bowlofeggs> here's an example of a package view in it: https://apps.fedoraproject.org/packages/bodhi 18:21:06 <bowlofeggs> it's actually a pretty handy app 18:21:24 <bowlofeggs> i use it a lot, primarily to find out which versions of a package are in which releases 18:21:41 <nirik> Oh, also bodhi uses it to complete package names 18:21:55 <nirik> and pkgwat command line tool uses it for all it's queries 18:22:10 <nirik> bowlofeggs: pkgwat releases bodhi :) 18:22:29 <bowlofeggs> i didn't even know about pkgwat 18:23:00 <cverna> python-appstream required to build fedora-packages not available in epel7 18:23:23 <nauseousSartre> oh, thanks. It makes sense now. 18:23:27 <cverna> that is where I stopped during the last apprentice work day 18:23:28 <nirik> cverna: Orig we didn't want to maintain TG in epel7, but someone is doing so now I guess. 18:23:29 <bowlofeggs> this might be a good opportunity to port it to flask? 18:23:48 <cverna> \o/ 18:23:56 <nirik> bowlofeggs: possibly. Might be less work, but might be more. Depends on people doing the work I guess 18:24:02 <bowlofeggs> yeah hard to guess 18:25:04 <nirik> icons seem broken too. ;( I was going to blow all your minds with 'pkgwat icon' :) 18:25:51 <nirik> anyhow, we need to fix it for all the above reasons. 18:26:11 <smooge> it was on a continual 'we will rewrite from scratch after X' 18:26:38 <nirik> it was orig "Fedora Community" 18:26:43 <nirik> and morphed into packages. 18:27:06 <nirik> (fedora community was kind of a early hubs thing... where maintainers could go to see a bunch of things) 18:27:10 <cverna> maybe we could set was is the minimun functionalities we currently need 18:27:20 <cverna> s/was/what 18:27:39 <cverna> and decide if a rewrite could be better 18:27:58 <nirik> well, short term if we could get it on rhel7, and fix the updating of things and bodhi package completion I think we could look at a re-write then? 18:28:11 <bowlofeggs> agreed 18:28:34 <nirik> it might not be too hard to remove pkgdb and replace with pdc querys 18:28:59 <threebean> some of this is done in git.. 18:28:59 <nirik> but I don't know. ;) 18:29:04 <threebean> (pkgdb vs pdc in fedora-packages) 18:29:13 <threebean> https://github.com/fedora-infra/fedora-packages/pulls 18:29:15 <nauseousSartre> nirik: sry for newbie question, but I was previously under the impresion that rhel doesn't use community solutions 18:29:40 <nirik> nauseousSartre: not sure what you mean there... 18:29:58 <nirik> Fedora Infrastructure uses only 100% open source stuff wherever possible 18:30:16 <nirik> threebean: ah ha 18:30:41 <nauseousSartre> alright, I was thinking that CentOS was the open-source version of rhel 18:31:09 <nauseousSartre> but I guess it's just the customer support then? 18:31:27 <nirik> yes, CentOS is all open source as well... (so thus similar to us) 18:31:41 <smooge> we just use RHEL in Fedora for historical reasons 18:31:57 <nirik> yeah. 18:32:02 <threebean> nirik: a problem -> no one seems to currently own merging/deploying/monitoring the app in production. 18:32:02 <bowlofeggs> RHEL is open source too! 18:32:20 <nirik> yes, true. :) 18:32:24 <nauseousSartre> smooge: so is the standard development environment for infrastructure RHEL? 18:32:40 <bowlofeggs> (the existence of CentOS proves the open-sourceness of RHEL :) ) 18:32:44 <nirik> threebean: yeah. :( 18:33:19 <smooge> nauseousSartre, it depends on the problem trying to be solved. We use Fedora for build servers and certain applications we can move easily and quickly every 6 months 18:33:23 <nirik> nauseousSartre: currently we deploy applications on Fedora or RHEL, but you can develop on Fedora, RHEL, CentOS, Scientific-linux, etc. 18:33:26 <bowlofeggs> nauseousSartre: i personally develop on fedora for bodhi, but i bet you could develop on RHEL or CentOS just as well 18:33:55 <nauseousSartre> thanks nirik and bowlofeggs 18:34:08 <nauseousSartre> and smooge 18:34:25 <nirik> nauseousSartre: sorry if I confused you... I didn't get the question was about rhel vs centos. :( 18:34:31 <smooge> anything else on this topic? bowlofeggs threebean do you have notes from FLOCK? 18:35:00 <nauseousSartre> nirik: nah, the question was confusing to begin with ;) 18:35:16 <nirik> I'm willing to look at building packages rpms and deploying if someone more developery would be interested in merging things, making sure it works and doing a release. ;) 18:35:51 <threebean> smooge: nothing in particular. lots of releng-related conversations about the modular release. 18:36:12 <bowlofeggs> smooge: i haven't written up my flock experience yet, but we did get some bodhi PRs submitted as a result! 18:36:16 <threebean> I guess I'll bring up the question of how to streamline deploying apps to openshift, but it's maybe a question for another day. 18:36:32 <cverna> I can help on the developery side of packages 18:36:50 <bowlofeggs> cverna++ 18:36:50 <zodbot> bowlofeggs: Karma for cverna changed to 4 (for the f26 release cycle): https://badges.fedoraproject.org/tags/cookie/any 18:36:58 <cverna> I will be nagging threebean tho :P 18:36:59 <nirik> thanks cverna! 18:37:08 * threebean disappears in a puff of smoke 18:37:10 <smooge> ok my notes were mostly aimed at EPEL or stuff I got for the x86_32 group so not for this meeting. 18:37:21 <smooge> ok next up 18:37:34 <smooge> #topic Apprentice Open office hours 18:37:36 <nirik> threebean: I think we can look at making a standard openshift app template/doc... might help 18:39:24 <cverna> maybe we could have an openshift section in the infra-docs 18:40:09 <nirik> yep. We should definitely. 18:40:37 <bowlofeggs> yeah that would be helpful 18:40:54 <bowlofeggs> i've been wondering if it would be painful or not to *partially* deploy bodhi on openshift 18:41:15 <bowlofeggs> meaning, the masher would probably not be on it, but the frontends and other backend could more easiler? 18:41:18 <bowlofeggs> *easily 18:41:22 <bowlofeggs> easiler... 18:41:33 <bowlofeggs> the masher might be hard to get in there? 18:41:33 <smooge> easiliest? 18:41:42 <smooge> I think we wandered into open floor 18:41:47 <bowlofeggs> yeah,s orry 18:41:49 <smooge> #topic Open Floor 18:41:58 <nirik> might be we should test/get fedmsg working before we go moving bodhi. 18:42:06 <bowlofeggs> true 18:42:18 * bowlofeggs throws jcline under the bus again 18:42:34 <bowlofeggs> we should clone jcline and name him jclone 18:42:57 <cverna> jclone++ 18:43:04 <jcline> bowlofeggs, the issue about sending messages you brought up at flock is the main blocker, I think 18:44:22 <bowlofeggs> yeah, but i think the active=True thing that patrick educated me about might help 18:44:24 <jcline> I have a mostly-functional publishing service, but it needs API polish and such. 18:44:55 <bowlofeggs> i think the active=True thing is conceptually pretty similar to your publishing service proposal actually 18:45:04 <bowlofeggs> or at least, by my understanding of each of them 18:45:56 <jcline> Sort of, but it uses pub/sub rather than using req/rep 18:46:07 <bowlofeggs> relrod, nirik: can pods in openshift open connections to containers in other pods? 18:46:17 <bowlofeggs> i would guess not 18:46:22 <nirik> I think no. 18:46:29 <relrod> good question :) but I think no 18:46:36 <nirik> unless there's a relay setup to expose that to the world 18:46:48 <bowlofeggs> but active=True i think would make the containers open outbound connections, which i think would work 18:47:11 <nirik> yes, as long as the other end is reachable... 18:47:14 <bowlofeggs> yeah 18:47:47 <bowlofeggs> so the combination of a publishing service and all container applications using active=True will probably be the way to go 18:48:15 <jcline> Er, well. active=True would go away with a publishing service. 18:48:37 <bowlofeggs> jcline: wouldn't the publishing service be unable to connect tot he containers as a subscriber? 18:48:41 <nirik> yeah, so thats a or... but we could do one easily now and look at the other later 18:49:02 <jcline> The publishing service doesn't connect to anything 18:49:09 <bowlofeggs> my understanding of active=True is that it switches the direction of the connection to where the publisher establishes the connection, rather than the subscriber 18:49:21 <bowlofeggs> ah ok 18:49:28 <bowlofeggs> i guess my understanding is incomplete :) 18:50:11 <nirik> well, the producers connect to the publishing service locally right? and then it listens for subscribers? 18:50:18 <nirik> or am I mixing it all up too? ;) 18:51:27 <jcline> nirik, no that's right 18:51:40 <bowlofeggs> cool 18:51:44 <bowlofeggs> i'm sure we can solve it :) 18:52:09 <jcline> Yes 18:52:20 <smooge> ok if we are done here.. 18:52:25 <bowlofeggs> sure 18:52:33 <nirik> so that wont work for openshift... unless we also expose the publisher in openshift as a endpoint/router 18:52:34 <nirik> ? 18:52:43 <nirik> sure, we can move on 18:52:47 <smooge> oh sorry 18:52:58 <smooge> there is nothing more than close the meeting 18:53:04 <bowlofeggs> we could have a special meetign about this later if we need to 18:53:21 <bowlofeggs> yeah we can close 18:53:34 <smooge> thank you all for coming today 18:53:37 <smooge> #endmeeting