16:02:37 <dustymabe> #startmeeting fedora_cloud_meeting
16:02:38 <zodbot> Meeting started Tue Dec  8 16:02:37 2020 UTC.
16:02:38 <zodbot> This meeting is logged and archived in a public location.
16:02:38 <zodbot> The chair is dustymabe. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:02:38 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
16:02:38 <zodbot> The meeting name has been set to 'fedora_cloud_meeting'
16:02:42 <dustymabe> #topic roll call
16:02:45 <cyberpear> .helo2
16:02:53 <cyberpear> .hello2
16:02:54 <zodbot> cyberpear: cyberpear 'James Cassell' <fedoraproject@cyberpear.com>
16:02:58 <roshi> .hello2
16:02:59 <zodbot> roshi: roshi 'Mike Ruckman' <roshi@mykolab.com>
16:03:00 <otubo> .hello2
16:03:02 <zodbot> otubo: otubo 'Eduardo Otubo' <eterrell@redhat.com>
16:03:05 * King_InuYasha waves
16:03:08 <King_InuYasha> .hello ngompa
16:03:09 <zodbot> King_InuYasha: ngompa 'Neal Gompa' <ngompa13@gmail.com>
16:03:19 <davdunc> .hello davdunc
16:03:19 <zodbot> davdunc: davdunc 'David Duncan' <davdunc@amazon.com>
16:03:32 <jdoss> .hello2
16:03:33 <zodbot> jdoss: jdoss 'Joe Doss' <joe@solidadmin.com>
16:04:31 <dustymabe> #chair cyberpear roshi otubo King_InuYasha davdunc jdoss
16:04:31 <zodbot> Current chairs: King_InuYasha cyberpear davdunc dustymabe jdoss otubo roshi
16:04:46 <dustymabe> #topic Action items from last meeting
16:04:55 <dustymabe> I don't see any action items from our last meeting
16:05:10 <dustymabe> #topic OpenStack cloud image: include and enable qemu-guest-image
16:05:19 <dustymabe> #link https://pagure.io/cloud-sig/issue/319
16:05:40 <dustymabe> we got a request to add the qemu-guest-agent to our cloud image
16:05:53 <dustymabe> looks like jdoss voiced some concern in the ticket
16:06:05 <jdoss> Is that just to add it or to start it for every provider?
16:06:33 <dustymabe> I'm not totally sure. I'm not sure if you can just include it and not have it enabled
16:06:50 <King_InuYasha> the systemd unit can be set to condition activation on kvm
16:06:51 * dustymabe vaguely remembers something about it being activated only if it would be useful for the machine
16:06:52 <jdoss> If it is just adding the package that isn't as bad as having a service running on GCP or AWS that is isn't needced.
16:07:14 <dustymabe> jdoss: at least for GCP we can remove it from the package set, because we have a separate GCP image
16:07:45 <dustymabe> King_InuYasha: that's good to know
16:07:50 <jdoss> The blanket add package and have the service running for everyone feels bad.
16:08:21 <dustymabe> jdoss: yeah, assuming it's only activated when useful, then that would be a little more palatable?
16:08:48 <jdoss> I don't think it's a big package so ya that works to help them with their usecase.
16:09:09 <jdoss> they should be able to enable it via userdata or something?
16:10:04 <dustymabe> King_InuYasha: is the condition you talk about part of the existing rpm?
16:10:10 <dustymabe> or were you talking about doing something extra?
16:10:24 <King_InuYasha> dustymabe: if it's not there now, we should add it
16:10:28 <King_InuYasha> I think it already is, though
16:11:14 <dustymabe> here's what I see: https://paste.centos.org/view/3aeedcad
16:12:04 <King_InuYasha> so it does it indirectly?
16:12:15 <King_InuYasha> but yeah, it doesn't have a virtualization condition
16:12:34 <dustymabe> King_InuYasha: because of the
16:12:34 <King_InuYasha> I think what that means is that if the device exists, it'll get started
16:12:42 <dustymabe> BindsTo=dev-virtio\x2dports-org.qemu.guest_agent.0.device
16:12:44 <dustymabe> After=dev-virtio\x2dports-org.qemu.guest_agent.0.device
16:12:54 <jdoss> That's how I am reading it.
16:12:56 <King_InuYasha> yeah
16:13:04 <King_InuYasha> I think we're fine to preload this unconditionally
16:13:10 <dustymabe> got ya, ok
16:13:12 <jdoss> sounds like we just need to add the package and move on with our lives.
16:13:20 <King_InuYasha> yup
16:13:22 <dustymabe> WFM
16:13:28 <davdunc> on the EC2 side, that would be considered a lot of overhead on the instance.
16:13:32 <cyberpear> maybe related: the cloud-init in openstack vs plain libvirt/qemu problem is annoying... otherwise, you could have the same image, I'd suppose... is it possible to have cloud-init be inert on plain qemu? (or have I just not tried it recently enough?)
16:13:54 <jdoss> davdunc: how so?
16:14:03 <dustymabe> cyberpear: it looks like it will be inert on any system that doesn't have dev-virtio\x2dports-org.qemu.guest_agent.0.device
16:14:05 <jdoss> would that device be present and start this service?
16:14:11 <jdoss> on EC2?
16:14:21 <cyberpear> interesting
16:14:30 <davdunc> jdoss, it would not, but it will affect boot time in the search.
16:15:31 <jdoss> a lot of overhead meaning a few MS to see if the dev is present or not? I can see what you are saying.
16:15:43 <davdunc> minimal, I know, but these add up.
16:15:46 <jdoss> ya for sure.
16:15:48 <michaelgugino> .hello2
16:15:50 <zodbot> michaelgugino: michaelgugino 'Michael Gugino' <gugino.michael@yahoo.com>
16:15:55 <jdoss> It's a good point.
16:16:00 * jdoss waves at Michael
16:17:00 <dustymabe> #proposed since the unit appears to be inert on any system that can't take advantage of the service, we propose to add qemu-guest-agent to the base package set. We won't include it on GCP since we have a separate image for GCP already. In the future we may make other images specific to cloud providers, in which case we'll also remove it from those variants.
16:17:43 <dustymabe> @chair michaelgugino
16:17:46 <King_InuYasha> dustymabe: +1
16:17:47 <dustymabe> sigh
16:17:50 <dustymabe> #chair michaelgugino
16:17:50 <zodbot> Current chairs: King_InuYasha cyberpear davdunc dustymabe jdoss michaelgugino otubo roshi
16:17:53 <davdunc> dustymabe: +1
16:18:43 <cyberpear> +1
16:19:14 <dustymabe> going once..
16:19:27 <danken> hu
16:19:39 <dustymabe> anybody else want to weigh in?
16:19:41 <danken> dustymabe: I arrived late
16:19:50 <dustymabe> welcome danken
16:20:09 <danken> are you adjourning?
16:20:17 <dustymabe> nope
16:20:23 <dustymabe> we're in the middle of a topic
16:20:33 <dustymabe> #agreed since the unit appears to be inert on any system that can't take advantage of the service, we propose to add qemu-guest-agent to the base package set. We won't include it on GCP since we have a separate image for GCP already. In the future we may make other images specific to cloud providers, in which case we'll also remove it from those variants.
16:21:33 <dustymabe> #topic open floor
16:21:45 <dustymabe> I don't think there are any other meeting tickets
16:22:08 <dustymabe> anybody with anything for open floor?
16:23:32 <dustymabe> oh, I almost forgot
16:23:36 * dustymabe checks calendar
16:23:56 <dustymabe> a few clerical items
16:24:02 <dustymabe> 1. the next scheduled meeting is on the 22nd
16:24:29 <dustymabe> I won't be around, and a lot of people will probably be AFK that time of year
16:25:03 <dustymabe> 2. I'm due to have a baby any time in the next 3 weeks. After that I won't be around for some time.
16:25:16 <dustymabe> Anybody else want to volunteer to run meetings (or rotate running meetings) in the meantime?
16:26:39 <michaelgugino> I could run some meetings
16:29:16 <danken> sorry for my earlier noise. is this a good time to raise my request to enable qemu-guest-agent in the Fedora cloud image?
16:29:44 <dustymabe> danken: :)
16:29:51 <danken> hmm
16:29:54 <dustymabe> we already discussed it and agreed to include it
16:30:06 <danken> did you also already discussed our request for ipv6-by-default?
16:30:06 <dustymabe> mostly because it appears to be inert on any nodes where it won't be useful
16:30:20 <dustymabe> danken: we did not discuss that one, was there a ticket opened for it?
16:30:33 <danken> I believe so, lemee check
16:31:00 <dustymabe> i may have missed it
16:31:20 <dustymabe> ahh I see, will add the meeting label
16:31:27 <danken> https://pagure.io/cloud-sig/issue/320
16:32:02 <dustymabe> #topic fedora cloud image: enable ipv6 (accept RAs by default)
16:32:08 <dustymabe> #link https://pagure.io/cloud-sig/issue/320
16:32:27 <dustymabe> It appears the network manager config in the cloud image sets `ipv6.method:                            ignore`
16:32:48 <dustymabe> honestly we should probably just stop configuring the network at all and just let NM defaults apply
16:33:02 <davdunc> that sounds right.
16:33:27 <dustymabe> anybody with any opinions here?
16:33:41 <danken> ... unless an NM default is later found to be buggy
16:34:09 <davdunc> can't hurt to build one for testing.
16:34:11 <danken> but to me - it would better be fixed in NM...
16:34:55 <dustymabe> yeah we should be able to apply it to rawhide and we'll get an image for testing
16:35:06 <dustymabe> danken: unfortunately we won't be able to update existing f33 images
16:35:11 <danken> davdunc: I'd love to consume a build with qemu-guest-agent and ipv6 and test it
16:35:18 <dustymabe> we only do a new release on major Fedora releases right now
16:35:50 <dustymabe> danken: what is your Fedora account nickname? is it mdbarroso ?
16:35:53 <danken> dustymabe: I see.
16:36:02 <danken> mine is danken...
16:36:29 <danken> but mdbarroso is in my team and should be CCed
16:37:29 <dustymabe> #proposed we should/will remove networking configuration from the cloud images and let NetworkManager default configs apply, which should get us in a situation where ipv6 is no longer ignored.
16:37:57 <davdunc> +1
16:38:55 <King_InuYasha> dustymabe: +1
16:39:14 <otubo> dustymabe: +1
16:39:52 <dustymabe> #agreed we should/will remove networking configuration from the cloud images and let NetworkManager default configs apply, which should get us in a situation where ipv6 is no longer ignored.
16:40:04 <dustymabe> #topic open floor
16:40:17 <dustymabe> anybody want to volunteer to run through making either of those two changes to the cloud base image?
16:41:08 <davdunc> dustymabe: I'd like to do it if I can get assistance from someone else, like @King_InuYasha
16:41:47 <davdunc> :D I need the experience.
16:42:38 <dustymabe> davdunc: I can help you :)
16:42:48 <davdunc> dustymabe: awesome!
16:42:59 <dustymabe> can you schedule some time (calendar invite) and I'll help you step through it?
16:43:09 <davdunc> will do.
16:43:32 <dustymabe> #action davdunc to make changes to the cloud base image to include qemu-guest-agent and drop the networking configuration
16:43:42 <dustymabe> anymore items for open floor?
16:46:22 <dustymabe> michaelgugino: maybe we can get updates on the image uploading bits in the new year?
16:46:42 <michaelgugino> dustymabe: for sure, updates.  I plan to hack on it soon.
16:47:03 <dustymabe> King_InuYasha: any progress with ericedens on the packaging front?
16:47:32 <dustymabe> also anyone want to volunteer to add the GCP links to the download page? https://pagure.io/cloud-sig/issue/314
16:48:03 <King_InuYasha> dustymabe: never heard back from him
16:48:05 <dustymabe> #info we have a PR open to rebase cloud-init to the new upstream version: https://src.fedoraproject.org/rpms/cloud-init/pull-request/11
16:48:33 <dustymabe> King_InuYasha: I have a meeting with him later today (if he's not out). Want me to add you ?
16:48:46 <King_InuYasha> sure
16:48:52 <King_InuYasha> depending on the time
16:48:56 <King_InuYasha> after 2pm
16:49:00 <King_InuYasha> (EST)
16:49:15 <dustymabe> starts at 2pm, is that OK
16:51:05 <dustymabe> ok I'll close out the meeting and coordinate with King_InuYasha outside of this
16:51:07 <dustymabe> #endmeeting