16:01:51 <mnaser> #startmeeting openstack_ansible_meeting
16:01:52 <openstack> Meeting started Tue Apr  9 16:01:51 2019 UTC and is due to finish in 60 minutes.  The chair is mnaser. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:01:53 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
16:01:56 <guilhermesp> will do it asap, thanks for the heads up jrosser
16:01:56 <openstack> The meeting name has been set to 'openstack_ansible_meeting'
16:01:57 <mnaser> #topic office hours
16:01:58 <guilhermesp> o?
16:02:01 <mnaser> noonedeadpunk[m]: yep!
16:02:03 <guilhermesp> o/*
16:03:38 <mnaser> reminder once again; https://etherpad.openstack.org/p/osa-train-ptg
16:04:33 <guilhermesp> probably a removal of centos support could be discussed too?
16:04:45 <guilhermesp> or do we keep centos at you distribution range?
16:04:57 <starborn> mnaser: Thanks. I'll try it this way :)
16:05:16 <mnaser> guilhermesp: I don't think we'll remove centos, there seem to be people who want to use it
16:05:21 <mnaser> starborn: good luck!
16:05:38 <mnaser> I think we'll keep it around as long as it doesn't become a time sink
16:06:09 <nicolasbock> o/
16:06:39 <noonedeadpunk[m]> And what about centos 8? Do we plan to support it as well?
16:06:54 <mnaser> noonedeadpunk[m]: we probably should.  centos 8 will allow us to bring python support
16:06:57 <mnaser> python 3 support*
16:07:07 <evrardjp> o/
16:07:42 <noonedeadpunk[m]> Agree
16:08:17 <jrosser> we'd have to tackle the lack of systemd-networkd
16:11:44 <cloudnull> o/
16:12:18 <cloudnull> ^ that's an issue, everything at this time in terms of networking for cent8 will require us to use network-manager
16:12:52 <cloudnull> all other network things that could be there will be removed.
16:13:51 <cloudnull> https://bugzilla.redhat.com/show_bug.cgi?id=1650342
16:13:51 <openstack> bugzilla.redhat.com bug 1650342 in systemd "systemd-networkd support in RHEL 8" [Unspecified,New] - Assigned to systemd-maint
16:14:06 <cloudnull> if anyone has rh access please pile on to that bug
16:15:06 <mnaser> can we just like
16:15:11 <cloudnull> looks like there's a new entry there for unofficial packages, maybe we could get infra to build cent8 images with that package pre-installed
16:15:13 <openstackgerrit> Guilherme  Steinmuller Pimentel proposed openstack/openstack-ansible master: debian: add job template for telemetry  https://review.openstack.org/651288
16:15:14 <mnaser> install a systemd-networkd rpm or something
16:15:27 <cloudnull> cent/rhel official repos wont have it
16:17:23 <mnaser> cloudnull: epel?
16:17:34 <cloudnull> maybe?
16:17:46 <cloudnull> if that is an option that would be great!
16:17:52 <cloudnull> then we dont have to change much, if anything
16:18:18 <cloudnull> its still too early to tell, still a lot of speculation at this poin t
16:18:32 <mnaser> yeah
16:18:36 <mnaser> who knows when RHEL 8 is gonna be out
16:18:42 <mnaser> and when centos will be out
16:18:43 <cloudnull> but it looks like others have chimed in so maybe there's still hope
16:18:50 <cloudnull> :D
16:19:25 <mnaser> cool cool
16:19:45 <mnaser> cloudnull: jrosser was suggesting if we can try to do something around deploying with nspawn but without macvlan
16:20:07 <cloudnull> sure
16:20:12 <cloudnull> we could do that
16:20:15 <mnaser> I think that will help us make it more of an 'achievable' target
16:20:50 <mnaser> all the stuff which are pooping out were mostly macvlan related at the time
16:20:53 <cloudnull> given its still "unstable" I would be game to reworking how we support it
16:21:01 <mnaser> yeah
16:21:35 <mnaser> I think debian-stable is really exciting
16:21:47 <mnaser> so that should be landing soon™
16:22:51 <cloudnull> :D
16:23:37 <evrardjp> cloudnull: long ago I suggested to do just only veth for having a smooth transition :)
16:24:00 <cloudnull> sadly, with nspawn, you cant do multiple veth
16:24:09 <evrardjp> oh yeah I remember that thing :)
16:24:16 <cloudnull> its a 1:1 :'(
16:24:24 <evrardjp> yeah I remember now
16:24:32 <cloudnull> however we could do host networking ?
16:24:35 <evrardjp> wan't this fixed?
16:24:36 <cloudnull> or ipvlan
16:25:09 <evrardjp> host networking would be simpler
16:25:32 <evrardjp> I kinda don't like it though
16:25:33 <jrosser> hmm? https://github.com/systemd/systemd/issues/3044
16:25:36 <cloudnull> there are new options in kernel 5.x which I would suspect nobody is running except us crazy people
16:26:05 <noonedeadpunk> cloudnull: btw, is it possible just to forward interface into nspawn? (asking in terms of IB interfaces - haven't tried this out yet)
16:26:25 <cloudnull> yes you can give an interface to a container
16:27:44 <mnaser> if we do host networking, is there much benefit to running in system containers now
16:27:57 <cloudnull> I would say no
16:28:05 <cloudnull> but, meh
16:29:00 <cloudnull> all that said, if folks have cycles, I'd love assistance on rethinking and making stable our nspawn solutions
16:30:25 <mnaser> cloudnull: is there a way to use systemd shenanigans to get services to run in isolated network?
16:30:34 <cloudnull> yes :)
16:30:40 <mnaser> I think this seems to be the biggest driver why folks like jrosser want to keep containers
16:30:49 <cloudnull> https://cloudnull.io/2019/04/running-services-in-network-name-spaces-with-systemd/
16:31:06 <cloudnull> which was mostly integrated into the systemd-service role
16:31:40 <cloudnull> https://github.com/openstack/ansible-role-systemd_service/blob/master/defaults/main.yml#L42-L68
16:32:24 <cloudnull> so my loadbancers in my dev cloud are running what my blog post covers
16:32:55 <cloudnull> which are totally isolated into a network namespace
16:33:38 <mnaser> mmm
16:33:45 <mnaser> need more time
16:33:47 <mnaser> :(
16:34:01 * mnaser has a lot of migrating to Debian ahead of us
16:34:53 <jrosser> cloudnull: what am i missing with --network-veth-extra=.....
16:41:09 <logan-> doesn't --network-veth-extra create all of the veths attached to a single bridge only
16:41:47 <logan-> https://github.com/systemd/systemd/issues/11087
16:50:37 <jrosser> "Takes a colon-separated pair of host interface name and container interface name."
16:59:18 <mnaser> #endmeeting