16:30:34 <lorbus> #startmeeting fedora_coreos_meeting
16:30:34 <zodbot> Meeting started Wed Oct 31 16:30:34 2018 UTC.
16:30:34 <zodbot> This meeting is logged and archived in a public location.
16:30:34 <zodbot> The chair is lorbus. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:30:34 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
16:30:34 <zodbot> The meeting name has been set to 'fedora_coreos_meeting'
16:30:44 <kaeso> .hello lucab
16:30:45 <zodbot> kaeso: lucab 'Luca Bruno' <lucab@redhat.com>
16:30:47 <lorbus> #topic Roll Call
16:30:49 <dustymabe> .hello2
16:30:50 <slowrie> .hello2
16:30:50 <ajeddeloh> .hello2
16:30:51 <zodbot> dustymabe: dustymabe 'Dusty Mabe' <dusty@dustymabe.com>
16:30:54 <zodbot> slowrie: slowrie 'Stephen Lowrie' <slowrie@redhat.com>
16:30:57 <zodbot> ajeddeloh: ajeddeloh 'Andrew Jeddeloh' <andrew.jeddeloh@redhat.com>
16:31:05 <bgilbert> .hello2
16:31:06 <zodbot> bgilbert: bgilbert 'Benjamin Gilbert' <bgilbert@backtick.net>
16:31:12 <sayan> .hello sayanchowdhury
16:31:13 <zodbot> sayan: sayanchowdhury 'Sayan Chowdhury' <sayan.chowdhury2012@gmail.com>
16:31:18 <jbrooks> .fas jasonbrooks
16:31:20 <zodbot> jbrooks: jasonbrooks 'Jason Brooks' <jbrooks@redhat.com>
16:31:23 <lorbus> #chair kaeso dustymabe slowrie bgilbert sayan jbrooks
16:31:23 <zodbot> Current chairs: bgilbert dustymabe jbrooks kaeso lorbus sayan slowrie
16:31:39 <mnguyen_> .hello mnguyen
16:31:39 <lorbus> #chair ajeddeloh
16:31:40 <zodbot> Current chairs: ajeddeloh bgilbert dustymabe jbrooks kaeso lorbus sayan slowrie
16:31:40 <zodbot> mnguyen_: mnguyen 'Michael Nguyen' <mnguyen@redhat.com>
16:31:53 <lorbus> #chair mnguyen_
16:31:53 <zodbot> Current chairs: ajeddeloh bgilbert dustymabe jbrooks kaeso lorbus mnguyen_ sayan slowrie
16:32:21 <lorbus> #topic Action items from last meeting
16:32:30 <lorbus> * bgilbert to PR the design doc and open tickets for backport-related questions
16:32:30 <lorbus> * ajeddeloh and lorbus to experiment on grubenv + static grub config
16:32:30 <lorbus> * kaeso to add summary of discussion with lennart re: portable services to #37
16:32:30 <lorbus> * dustymabe to open tickets for individual clouds so that we can document gaps/strategy for not shipping cloud agents (#12)
16:34:02 <ajeddeloh> lorbus: I haven't had time to work on that, dunno about you
16:34:22 * lorbus neither
16:34:46 <lorbus> I saw that kaeso added a comment: https://github.com/coreos/fedora-coreos-tracker/issues/37#issuecomment-433889365
16:34:56 <kaeso> ignition+portablectl summary is on GH
16:34:57 <kaeso> yup
16:35:01 <dustymabe> #info dustymabe opened tickets #65 -> #71 for cloud agent investigation
16:35:01 <kaeso> #link https://github.com/coreos/fedora-coreos-tracker/issues/37#issuecomment-433889365
16:35:02 <lorbus> so that's done :)
16:35:32 <lorbus> bgilbert, do you have an update, too?
16:35:50 <dustymabe> lorbus: want to re-action the grubenv item?
16:36:06 <lorbus> dustymabe: sure
16:36:10 <bgilbert> yup, the PR is up in https://github.com/coreos/fedora-coreos-tracker/pull/72
16:36:21 <bgilbert> started working on the bugs, then went down a rabbit hole of bug-writing
16:36:28 <bgilbert> those should be up later today
16:36:38 <lorbus> #action ajeddeloh and lorbus to experiment on grubenv + static grub config
16:36:58 <dustymabe> #info bgilbert opened PR for design doc for release streams discussion decision
16:37:02 <bgilbert> #link https://github.com/coreos/fedora-coreos-tracker/pull/72
16:37:22 <lorbus> dustymabe's typing faster
16:37:31 <rfairley> .hello rfairleyredhat
16:37:32 <zodbot> rfairley: rfairleyredhat 'Robert Fairley' <rfairley@redhat.com>
16:37:34 * dustymabe rests keyboard
16:37:42 <lorbus> #chair rfairley
16:37:42 <zodbot> Current chairs: ajeddeloh bgilbert dustymabe jbrooks kaeso lorbus mnguyen_ rfairley sayan slowrie
16:38:03 <lorbus> than me I should've noted :P damn US keyboard
16:38:38 <lorbus> #topic Cloud Agents
16:38:50 <lorbus> dustymabe go
16:39:07 <slowrie> it might make more sense to dive in on a per cloud basis
16:39:42 <lorbus> slowrie: sure. any specific one you have in mind already?
16:40:39 <slowrie> The main ones we probably want discussion around are probably going to be GCE, Azure & VMware
16:40:48 <dustymabe> sounds good to me
16:40:57 <dustymabe> want to go in that order?
16:41:12 <slowrie> sgtm, ajeddeloh want to chime in on the oslogin stuff?
16:41:21 <lorbus> #topic GCE Cloud Agent
16:41:26 <ajeddeloh> sure
16:41:41 <ajeddeloh> so first, there's more than just oslogin to be on google cloud
16:41:58 <dustymabe> #link https://github.com/coreos/fedora-coreos-tracker/issues/67
16:41:59 <ajeddeloh> but I think everything else can be containerized like we do on CL
16:42:22 <lorbus> dustymabe: damn you, again^^
16:42:32 <ajeddeloh> oslogin is basically just a couple pam and nss modules
16:42:40 <ajeddeloh> and the config files to enable them
16:42:50 <ajeddeloh> oh and an authorizedSSHKeys binary
16:43:33 <ajeddeloh> assuming we're shipping the same image everywhere, that'll mean we'll ship said modules everywhere but only enable them via configuration elsewhere
16:43:39 <dustymabe> ajeddeloh: so would we just embed it in the ostree and conditionally enable oslogin?
16:43:45 <ajeddeloh> yeah
16:43:49 <dustymabe> got ya
16:44:14 <dustymabe> would we propose to create a google-oslogin rpm in fedora then?
16:44:14 <lorbus> anything else here?
16:44:27 <ajeddeloh> And I think we want the ability to disable them if the user wants (via Ignition)
16:44:33 <dustymabe> i know there are some that exist but I don't think any in fedora
16:44:47 <ajeddeloh> +1 yeah that makes sense
16:44:54 <kaeso> there a few more non-agent bits on GCP
16:44:56 <slowrie> from the ticket it looks like that's already happening dustymabe
16:44:59 * dustymabe assumes it's open source
16:45:05 <ajeddeloh> it is
16:45:20 <dustymabe> what is alredy happening ?
16:45:25 <ajeddeloh> not _too_ bad to package (got a few quirks, otherwise decent, not too many deps)
16:45:31 <slowrie> the creation of a google-oslogin rpm
16:45:36 <slowrie> see jdoss' comment
16:45:53 <ajeddeloh> one other thing
16:46:00 <kaeso> I know that on CL we do something to containerize gsutil
16:46:18 <ajeddeloh> on non-CL distros OSLogin works a little differently
16:46:18 <dustymabe> hmm. his comment links to an rpm
16:46:29 <dustymabe> i don't know of any specific effor to get that rpm into fedora though
16:46:36 <slowrie> ah
16:46:37 <dustymabe> we could ask to see if he's interested
16:46:38 <slowrie> Might have misread
16:46:49 <jdoss> dustymabe: https://pagure.io/fedora-server/issue/5#comment-538460
16:46:59 <ajeddeloh> on "normal" OSs the gce agent is the thing enabling/disabling oslogin
16:47:21 <dustymabe> oh nice
16:47:23 <jdoss> ^ Current effort IIRC I have not used it for my GCP deployments
16:47:29 <dustymabe> so we just need to finish the package reviews
16:47:31 <ajeddeloh> and it gets a signal from gce to toggle it (so you can toggle it in one place on the cloud side)
16:47:58 <jdoss> .hello2
16:47:59 <zodbot> jdoss: jdoss 'Joe Doss' <joe@solidadmin.com>
16:48:05 <lorbus> #chair jdoss
16:48:05 <zodbot> Current chairs: ajeddeloh bgilbert dustymabe jbrooks jdoss kaeso lorbus mnguyen_ rfairley sayan slowrie
16:48:07 <ajeddeloh> but that's via some bash that manually mangles things like nsswitch.conf
16:48:20 <dustymabe> ok so what are the non-agent bits needed for GCP ?
16:49:07 <slowrie> Not sure if gsutil is required but we might want to talk about the inclusion of gsutil/awscli
16:49:16 * ajeddeloh doesn't remember all of them off the top of his head
16:49:25 <ajeddeloh> but there's something for networking, time
16:49:32 <kaeso> I was digging in GH tracker
16:49:44 <kaeso> and we were aiming at just removing that
16:49:47 <ajeddeloh> awscli containerizes nicely, yes?
16:49:51 <kaeso> #link https://github.com/coreos/bugs/issues/2235#issuecomment-344749342
16:49:54 <slowrie> believe it should
16:50:17 <jdoss> gcloud / gsutil should too
16:50:31 <bgilbert> well, mostly: https://github.com/coreos/bugs/issues/2235
16:51:00 <kaeso> I'm pretty sure there were other issues too, but I can't find them now
16:51:28 <dustymabe> yeah maybe we can open a separate issue for gsutil/awscli (cloud clients)
16:51:33 <dustymabe> slowrie: ^^
16:51:48 <slowrie> works for me
16:52:27 <lorbus> should we action somebody to do it?
16:52:29 <dustymabe> sound like an action item
16:52:31 <dustymabe> yep
16:52:45 <lorbus> any volunteers?
16:52:54 <kaeso> I'd be for just leaving those to the user, as normal application/utilities
16:53:10 <dustymabe> i nominate slowrie :)
16:53:14 <slowrie> I'm fine doing it
16:53:38 <kaeso> (as in, to run in their container runtime of choice)
16:53:39 <lorbus> #action slowrie to open separate issues for gsutil/awscli cloud clients
16:54:16 <dustymabe> ok regarding package reviews
16:54:42 <dustymabe> anybody interested in doing a package review https://pagure.io/fedora-server/issue/5#comment-538460 ?
16:56:04 <dustymabe> which would help us move things along.
16:56:19 <lorbus> #action lorbus to review gce-oslogin rpm https://pagure.io/fedora-server/issue/5#comment-538460
16:56:23 <dustymabe> I'm not a sponsor yet, but I'm working on it
16:56:27 <dustymabe> thanks lorbus
16:56:28 <lorbus> I'll do it
16:56:48 <lorbus> anything else for GCE?
16:57:20 <lorbus> #topic Azure Cloud Agent
16:57:20 <lorbus> #link https://github.com/coreos/fedora-coreos-tracker/issues/65
16:57:34 <slowrie> I posted a bash script & systemd unit for the checkin in https://github.com/coreos/fedora-coreos-tracker/issues/65#issuecomment-434449560
16:58:17 <slowrie> It assumes the known wireserver IP address, which should always be the case for ARM machines so only ASM & AzureStack should have custom addrs
16:58:40 <kaeso> slowrie: have you seen https://github.com/coreos/coreos-metadata/issues/120 too?
16:59:12 <slowrie> Yeah, that ticket is in reference to where the final version should live
16:59:47 <slowrie> As it needs to run every boot it should live in coreos-metadata, that script is more of a PoC to see what is required to replace waagent
17:00:17 <dustymabe> +1
17:00:18 <slowrie> One currently outstanding issue is that there are no nameservers setup post-boot despite them being sent in the DHCP broadcasts
17:00:29 <dustymabe> do we think that is a networking bug ?
17:00:58 <slowrie> I'm not sure yet, I need to ensure that it is present in the initramfs and if so it's probably a networking bug
17:01:16 <bgilbert> there's probably some other agent functionality we'll need to look into as well; the Azure folks mentioned ephemeral disks
17:01:42 <dustymabe> bgilbert: like formatting and mounting them?
17:02:17 <bgilbert> at a guess.  I don't know any more yet
17:02:40 <dustymabe> +1
17:03:24 <lorbus> who'd like to pick up research on this?
17:03:41 <slowrie> I'm going to keep digging on the networking portion for now
17:03:45 <lorbus> slowrie again?
17:03:52 <dustymabe> yep. thanks slowrie
17:04:01 <dustymabe> we don't necessarily need any action items here
17:04:09 <dustymabe> since it's more open ended
17:04:15 <lorbus> ack
17:04:30 <slowrie> bgilbert mind updating the ticket w/ the ephemeral disks info so it doesn't slip off the radar?
17:05:13 <bgilbert> slowrie: I'll follow up with the Azure folks first.  want to make sure we have a solid understanding of what's needed, rather than fragments
17:05:19 <slowrie> +1
17:05:28 <lorbus> moving on, or anything else to add here?
17:05:45 <lorbus> #topic VMware Cloud Agent
17:05:51 <lorbus> #link https://github.com/coreos/fedora-coreos-tracker/issues/70
17:06:32 <slowrie> VMware is probably one of the agents we're going to just have to include
17:06:42 <slowrie> from my memory of past discussions the agent doesn't containerize nicely
17:06:48 <kaeso> that one IIRC is nasty because it's basically a configuration manager
17:07:51 <dustymabe> right. but what are the bare minimum features we actually need?
17:07:51 <jdoss> Do we have an issue for Virtualbox tools for people that want to run Fedora CoreOS in Vagrant with Virtualbox? I know if my dev coworkers don't have the tools installed, time drifts when suspended on OS X and it causes issues.
17:08:07 <jdoss> (maybe this isn't in scope but it's an agent I think)
17:08:09 <dustymabe> jdoss: yep https://github.com/coreos/fedora-coreos-tracker/issues/73
17:08:32 <dustymabe> kaeso: IOW could we not ship the agent and just implement the bare minimum features
17:08:47 <jdoss> +1 sorry for being behind :)
17:09:21 <kaeso> dustymabe: I don't know. I guess we can start without and start recording things that are missing
17:09:41 <dustymabe> yeah. i think that's the point of all of these tickets
17:09:55 <dustymabe> i.e. if we try not to ship the guest agent how much pain are we going to have to endure
17:10:06 <slowrie> iirc virtualbox outside of vagrant would require either a nasty UX or changes being made to virtualbox itself to support larger metadata sizes
17:10:29 <slowrie> bgilbert or kaeso might have more context
17:10:48 <kaeso> dustymabe: vmware is the one we are the least familiar I think
17:11:12 <bgilbert> slowrie: yeah, more in the ticket ^
17:11:56 <dustymabe> kaeso: IOW we'd need to get some hardware and investigate
17:12:25 <dustymabe> hardware == (set up environment)
17:13:20 <slowrie> dustymabe: we can probably test on packet's ESX servers, don't think we need vSphere for basic testing w/out the agent
17:13:34 <dustymabe> k
17:13:40 <dustymabe> i'll add a note to the ticket
17:13:55 <lorbus> anything I should info for this topic?
17:14:47 <lorbus> There is one more item for this week from our issue tracker: https://github.com/coreos/fedora-coreos-tracker/labels/meeting
17:14:55 <lorbus> #topic Docker Version
17:15:02 <lorbus> #link https://github.com/coreos/fedora-coreos-tracker/issues/64
17:15:23 <dustymabe> yep. just wanted to bring this up as a user opened the ticket
17:15:30 <dustymabe> and I wanted to make sure we are all on the same page
17:16:06 <dustymabe> anybody object to the responses I made in the ticket ?
17:16:50 <lorbus> #info FCOS will default to the moby-engine package which follows upstream
17:17:09 <kaeso> that seems correct
17:17:09 <lorbus> everybody ok with that?
17:17:23 <lorbus> #topic Open Floor
17:17:35 <kaeso> the open questions were related to "which version" and "how to switch version"
17:17:41 <kaeso> (modularity and overlaying?)
17:18:26 <dustymabe> kaeso: right. "which version" == whatever is in moby-engine package in fedora
17:18:36 <dustymabe> "how to switch" we need another issue for that I think
17:18:54 <dustymabe> but yes, modularity and overlaying was what I was thinking
17:18:58 <dustymabe> though will require some work
17:18:58 <kaeso> ack. So that ticket is technically correct :)
17:19:30 <dustymabe> anybody want to volunteer to add the moby-engine package to FCOS configs ?
17:19:39 <dustymabe> that could be a first step anyway
17:20:03 <lorbus> I'll do it
17:20:29 <lorbus> #action lorbus to add moby-engine package to FCOS configs
17:20:38 <kaeso> if we add it to base FCOS, can users overlay a different version later?
17:21:22 <jbrooks> yes
17:21:36 <dustymabe> yeah, but we'd need to make things "nicer" in the future
17:22:03 <dustymabe> this is mainly just a first step and us making a statement that we are including it
17:22:06 <jdoss> Why not just use podman to run moby-engine in a container ??? ;)
17:22:14 <kaeso> ack (it looks like I misunderstood this in the past)
17:22:22 * jdoss is so sorry for the bad joke
17:22:54 <lorbus> anybody got anything for open floor?
17:22:59 <dustymabe> lorbus: i do
17:23:00 <kaeso> jdoss: silly as it seems, we actually tried several iterations of that with systemd/rkt/docker and turns out it doesn't really work :)
17:23:15 <dustymabe> #info fedora 29 atomic host was released yesterday
17:23:26 <sayan> ksinny++
17:23:26 <zodbot> sayan: Karma for sinnykumari changed to 1 (for the current release cycle):  https://badges.fedoraproject.org/tags/cookie/any
17:23:28 <jdoss> haha, you all get points for trying kaeso
17:23:31 <lorbus> dustymabe: yay!
17:23:38 <dustymabe> which isn't super relevant for Fedora CoreOS, other than to explicitly call out that I expect FCOS development to pick up now
17:23:39 <lorbus> ksinny++
17:23:40 <zodbot> lorbus: Karma for sinnykumari changed to 2 (for the current release cycle):  https://badges.fedoraproject.org/tags/cookie/any
17:24:13 <dustymabe> mainly fedora 29 is out the door and fedora 30 is the cycle we are targeting for FCOS, so we're going to race to our target
17:24:40 <lorbus> fasten the seatbelts everyone!
17:24:44 <lorbus> :)
17:24:52 <dustymabe> also note that big changes we want to make or have happen in other parts of fedora (systemd networkmanager....) need to happen earlier iin the cycle than later
17:25:11 <dustymabe> earlier in the cycle means that we have a greater chance of it making whatever devcut the teams have for fedora 30
17:25:15 <jdoss> killing networkmanager and moving to systemd-networkd?
17:25:19 <jdoss> one can dream
17:25:29 <kaeso> *cough*pam*cough*
17:25:41 <dustymabe> so if you're working on something that involves other subsystems, working on that before other things might be a good idea
17:25:50 <bgilbert> kaeso: killing pam and moving to systemd-networkd?
17:26:02 <dustymabe> what did she ever do to you?
17:26:08 <dustymabe> pam's a nice lady
17:26:09 <jdoss> Jim is going to be so mad
17:26:14 <kaeso> that's a trip, not a dream :)
17:26:23 <dustymabe> :)
17:26:34 <dustymabe> i'm trying to think of any other open floor items
17:26:36 <dustymabe> anyone else?
17:27:03 <kaeso> the above was https://github.com/linux-pam/linux-pam/pull/69 bts
17:27:05 <kaeso> *btw
17:27:43 <dustymabe> kaeso: calling out that it hasn't been reviewed yet ?
17:27:58 <kaeso> :)
17:28:04 <lorbus> rfairleyredhat++
17:28:04 <zodbot> lorbus: Karma for rfairleyredhat changed to 1 (for the current release cycle):  https://badges.fedoraproject.org/tags/cookie/any
17:28:53 <mskarbek> in terms of docker/podman I'm working on podman wrapper (varlink based) which will expose docker api with hope that this will kill need for docker/moby for good
17:29:41 <jdoss> That sounds awesome mskarbek
17:30:10 <lorbus> #info mskarbek working on a varlink based podman wrapper to expose docker api
17:30:16 <dustymabe> mskarbek: nice
17:30:28 <dustymabe> is that something you've shared with upstream podman maintainers ?
17:30:34 <dustymabe> in case they are working on something similar?
17:30:44 <mskarbek> this is why I created docker version issue in the first place to make sure that I can focus on newer api versions
17:31:22 <lorbus> it's time!
17:31:27 <dustymabe> yep
17:31:31 <lorbus> thank you all for coming!
17:31:42 <lorbus> #endmeeting