14:03:19 <mattdm> #startmeeting Cloud WG (2014-07-10)
14:03:19 <zodbot> Meeting started Thu Jul 10 14:03:19 2014 UTC.  The chair is mattdm. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:03:19 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
14:03:23 <mattdm> #meetingname cloud
14:03:23 <zodbot> The meeting name has been set to 'cloud'
14:03:31 <mattdm> #topic who we got?
14:03:38 <mattdm> I'm here :)
14:03:57 <mattdm> anyone else?
14:04:13 <roshi> .hellomynameis
14:04:13 <zodbot> roshi: (hellomynameis <username>) -- Return brief information about a Fedora Account System username. Useful for things like meeting roll call and calling attention to yourself.
14:04:18 <roshi> .hellomynameis roshi
14:04:19 <zodbot> roshi: roshi 'Mike Ruckman' <mruckman@redhat.com>
14:04:30 <mattdm> .hellomynameis mattdm
14:04:31 <zodbot> mattdm: mattdm 'Matthew Miller' <mattdm@mattdm.org>
14:05:06 <mattdm> I guess we'll give it a few minutes. for some reason 10am seems extra early this morning
14:05:30 <roshi> well, it's 8 for me
14:05:44 <mattdm> see? crazy! :)
14:06:04 <roshi> timezones are weird :)
14:06:07 <roshi> how's the future?
14:06:11 <number80> .hellomynameis hguemar
14:06:12 <zodbot> number80: hguemar 'Haïkel Guémar' <karlthered@gmail.com>
14:06:14 <mattdm> sleepy!
14:06:35 <mattdm> welcome number80!
14:07:10 <mattdm> I don't know if we're going to get many more.... :(
14:07:38 <mattdm> The main topics are F21 Alpha related...
14:08:12 <mattdm> We should check in on all of the feature progress.... I guess we can do that on the list
14:08:20 <roshi> oh, mattdm - next blocker review will be an hour earlier - so fesco members can be there for the first hour or so :)
14:08:41 <mattdm> roshi cool. :)
14:08:50 <agrimm> .hellomynameis arg
14:08:51 <zodbot> agrimm: arg 'Andy Grimm' <agrimm@redhat.com>
14:09:08 * agrimm hates dueling meetings
14:09:09 <mattdm> from that, I see that there's an selinux bug which keeps anything from working?
14:09:34 <roshi> or a bug in the labeling of things - it wasn't that clear in my head
14:09:38 <mattdm> agrimm maybe it's time to reschedule this one again for people's summer schedule
14:10:09 <jsmith> .heelomynameis jsmith
14:10:12 <mattdm> #topic F21 TC1 and the selinux bug
14:10:13 * jsmith fails
14:10:39 <mattdm> the selinux bug is
14:10:41 <mattdm> #link https://bugzilla.redhat.com/show_bug.cgi?id=1116450
14:11:00 <mattdm> as I understand it, this will make the images not boot
14:11:15 <mattdm> so TC1 will be pointless for cloud (since there's no way to change boot params at runtime)
14:11:58 <mattdm> or is this not the case for ssh logins?
14:11:59 <imcleod> Sorry.  Had my head in the clouds....
14:12:15 <roshi> I think it was causing issues with sddm and ssh as well
14:12:19 * roshi looks for the bug
14:12:27 <jsmith> Yeah, I'm pretty sure it affects SSH as well
14:12:30 <mattdm> roshi see link above :)
14:12:57 <number80> I don't see how it would enable ssh logins but blocking tty logins :/
14:12:58 <roshi> https://bugzilla.redhat.com/show_bug.cgi?id=1052317
14:13:14 <mattdm> okay, so, I guess the basic choice is: a) wait on cloud testing until that's fixed or b) temporarily disable selinux on the command line for the images
14:13:27 <mattdm> any preferences?
14:13:43 <roshi> disabling selinux violates another release criteria
14:13:57 <roshi> it would 'work' though
14:14:24 <number80> could we relax it for TC1 so we could test other stuff ?
14:14:30 <imcleod> http://stopdisablingselinux.com/
14:14:35 <mattdm> imcleod lol
14:14:42 <jzb> argh, hi all
14:14:50 <mattdm> right I'm quite sure we wouldn't leave it disabled
14:14:52 <imcleod> I can already see the tears welling up in Dan's eyes.
14:14:54 <mattdm> jzb: good morning!
14:15:01 <number80> imcleod: I saw Dan disabling SELinux on his laptop (sorry bro, I can't unsee it)
14:15:02 <agrimm> mattdm, can we just do some magic in the kickstart to fix the labels that are broken?
14:15:06 <roshi> well, the thing with selinux I've found is - just becuase your tests pass with it disabled != they pass with it enabled
14:15:08 <jzb> mattdm: sorry, I have a hard conflict from now on at this time.
14:15:14 <mattdm> agrimm oh. um, yes, probably.
14:15:27 <agrimm> mattdm, so let's try that.  :)
14:15:32 <imcleod> mattdm, agrimm: We already have some selinux fixup in the default ks IIRC.
14:15:38 <mattdm> agrimm yes that is *way* better
14:15:42 <roshi> yeah - if there's a workaround, that would be better than disabling selinux
14:16:14 <imcleod> https://git.fedorahosted.org/cgit/spin-kickstarts.git/tree/fedora-cloud-base.ks#n219
14:16:18 <mattdm> jzb: let's look at rescheduling the meeting again. looks like it's not working for a lot of people
14:16:33 <jzb> mattdm: sending out the thingy right after this
14:16:39 <mattdm> jzb: cool
14:17:09 <mattdm> imcleod: oh yeah. one of my todo items is to look if that is still necessary. probably not, because it was appliance-creator hacks
14:17:40 <mattdm> but anyway: does anyone want to volunteer to find the correct hack and send it to me? I'll put it in the spin kickstart
14:17:54 * roshi doesn't have enough experience with kickstarts
14:18:19 <agrimm> mattdm, I can try, though I may run out of time
14:18:41 <mattdm> roshi: the part in question is just a bash script that runs after the install is complete
14:18:48 <number80> roshi: RH's manuals are good references btw for kickstart options
14:19:02 <mattdm> number80: yes but right now, not needed. :)
14:19:13 <roshi> I've been meaning to do some more research on it - just haven't had time
14:19:40 <mattdm> #action agrimm to try to figure out workaround for https://bugzilla.redhat.com/show_bug.cgi?id=1052317 for cloud kickstarts until bug is fixed
14:19:49 <mattdm> #info agrimm is overloaded though :)
14:20:05 <mattdm> anyway that seems like the right way forward....
14:20:39 <mattdm> the next topic is -- state of our changes.
14:21:03 <mattdm> #topic review F21 cloud changes
14:21:07 <mattdm> #link https://fedorahosted.org/cloud/ticket/72
14:21:25 <mattdm> #link https://fedoraproject.org/wiki/Cloud_Changelist
14:21:34 <mattdm> should we just go through the list?
14:21:50 <mattdm> it's kind of a big list.
14:22:28 <roshi> could probably start with reassigning owners since some seats changed
14:22:51 <number80> +1
14:23:19 <roshi> just a thought :) something low hanging
14:23:20 <mattdm> anyone opposed to the idea of going through the list? It's kind of a big list but most will be quick
14:23:34 <mattdm> a bit tedious but will make sure we're not dropping something
14:23:43 <imcleod> mattdm: I'll try to stick with you....
14:23:57 <mattdm> okay let's do it :)
14:24:09 <mattdm> #topic External Need: Automatic Image Upload
14:24:22 <mattdm> #info oddshocks has this well underway and it's going into production RSN
14:24:30 <mattdm> so basically, status is good
14:24:36 <roshi> real soon now?
14:24:50 <mattdm> roshi: yes. :)
14:25:07 <mattdm> it's functional, just needs to get into place in the infrastructure
14:25:21 <imcleod> Any external cloud other than EC2?
14:25:35 <imcleod> I assume not, the two I am involved with are still waiting on me to move the legal conversation along.
14:25:46 <mattdm> imcleod: right now I think it's just EC2 but I know he's been working on putting the parts in place pending the things you are working on
14:25:56 <imcleod> mattdm: Roger.
14:26:13 <mattdm> #topic External Need: Updated Web Site
14:26:42 <mattdm> #info websites team is working on this, looks to be making good progress
14:26:46 <mattdm> #link https://fedorahosted.org/fedora-websites/ticket/267
14:27:00 <mattdm> I don't think there is much more to say here :)
14:27:11 <mattdm> #topic External Need: Software Collections for Cloud Users
14:27:22 <mattdm> okay, so.... this one is dead in the water :(
14:27:26 <number80> sigh
14:27:37 <number80> I think that we should escalate this to fesco
14:27:57 <mattdm> FPC and the SCL developers are still working things out, and marcela has withdrawn the ruby scl
14:28:11 <mattdm> number80: possibly. what's the hoped-for goal?
14:28:20 <mattdm> of escalation, I mean.
14:28:43 <number80> to see how we could help FPC to speed things ups
14:29:10 <mattdm> from the sidelines, it looks like part of the slowness is due to global communications requiring a lot of back and forth...
14:29:30 <mattdm> but also some fundamental disagreements between devs and FPC
14:29:46 <number80> probably, organizing a meeting at Flock could help too
14:29:51 <mattdm> to me, it looked like the last thing came down to a _really trivial_ point about package naming that could have gone either way
14:30:09 <mattdm> number80 yeah, an in-person meeting might really help. but that'll be too late for F21
14:30:35 <mattdm> I think unless someone else wants to make a "showcase" SCL for F21, we're basically going to ship without it
14:30:36 <number80> agreed, but I think it's still realistic for F22
14:30:41 <mattdm> number80 *nod*
14:31:10 <mattdm> so my thinking is that we should really focus on Atomic as the Shiny Thing for F21
14:31:19 <mattdm> (which will come up below)
14:32:10 <mattdm> in some ways, Docker presents a different way of doing the same basic thing -- selected stack regardless of core
14:32:27 * jsmith tends to agree
14:33:05 <imcleod> I'd go so far as to say that Docker has benefited from our inability to provide the same thing within a single userspace....
14:33:18 <imcleod> Benefited quite a bit...  But that's not really on topic....
14:33:23 <mattdm> so, presuming the Atomic stuff stays on track (discussion coming soon), I think we should put that at the top of our presentation of Fedora Cloud, with the generic image below that as the "conservative fallback"
14:34:03 <mattdm> #info proposal for cloud wg (since we don't have quorum here): focus marketing efforts on Atomic
14:34:19 <mattdm> #topic External Need: Batched Updates
14:34:26 <mattdm> ooh this is easy
14:34:32 <mattdm> #info no one is working on this. still would be nice.
14:34:46 <mattdm> next :)
14:34:55 <mattdm> #topic External Need: Automatic Smoketests on Image Build
14:35:03 <mattdm> ah, this one is also easy, but in a good way
14:35:25 <mattdm> #link https://github.com/fedora-infra/fedimg/issues/7
14:35:30 <mattdm> #info this is implemented as part of fedimg
14:35:50 <mattdm> #info fancier notifications and taskotron integration remain on the todo list
14:36:09 <mattdm> #topic External Need: Scratch Builds on Change
14:36:41 <mattdm> #info this remains a dream, basically. but we're getting nightlies, so that's a good start
14:36:59 <mattdm> okay so now we're to the actual changes. anyone still with me? :)
14:37:06 <mattdm> #topic Change: Move to ImageFactory For image Creation
14:37:17 <imcleod> Done
14:37:23 <mattdm> #info Done!
14:37:25 <mattdm> yes :)
14:37:33 <mattdm> #topic Change: (A)Periodic Updates to the Images
14:37:46 <mattdm> #undo
14:37:46 <zodbot> Removing item from minutes: <MeetBot.items.Topic object at 0x1f872210>
14:37:56 <mattdm> #topic Change: (A)Periodic Updates to Cloud Images
14:38:09 <mattdm> Sooo... this one needs a driver.
14:38:28 <mattdm> I've poked at a few things but haven't really done a great job.
14:39:03 <mattdm> #link https://fedoraproject.org/wiki/Changes/(A)Periodic_Updates_to_Cloud_Images
14:39:06 <imcleod> So, if we have nightly builds, and automated uploads, isn't this almost trivial to implement?
14:39:11 <number80> It requires interacting with releng folks
14:39:20 <mattdm> see the "detailed description" for some details
14:39:29 <imcleod> Sorry.  Looking.
14:39:42 <mattdm> imcleod: yes, in a practical standpoint that's an important enabler
14:39:50 <mattdm> but we still need the policies and the qa expectations
14:40:18 <mattdm> in reality, automatic builds with smoketests are a cut *above* the ad hoc procedure we have now
14:40:19 <imcleod> mattdm: Right.  I am understating the significance of the policy questions.  As well as mirroring and QA...
14:40:36 <mattdm> but if we start doing it regularly (as planned), it sets the expecations higher
14:40:58 <mattdm> let's do this....
14:41:23 <mattdm> #help this really needs someone to drive -- who can commit?
14:41:35 <mattdm> #action mattdm to ask that question on the cloud mailing list
14:41:59 <mattdm> #topic Atomic Cloud Image
14:42:08 <mattdm> #info this one needs a new owner
14:42:22 <mattdm> sandro had this one so we need a replacement
14:43:23 <mattdm> jzb I know you're working with Atomic upstream... is this something that fits into what you can take on?
14:43:41 * mattdm waits because I know jzb is in another meeting
14:44:12 <mattdm> (note that Colin is already a co-owner so much of the ostree-related technical side is already covered)
14:44:58 <mattdm> hmmm maybe I'll take this to the list as well :)
14:45:28 <mattdm> #help Atomic Cloud Image needs a new co-owner/coordinator
14:45:42 <mattdm> #action mattdm to ask about atomic change owner on list
14:45:57 <mattdm> #topic Change: Convert Fedora Cloud Image to Fedora Server ("Adopt Your Cattle")
14:46:21 <mattdm> #info mattdm and sgallagh own this. it isn't started yet, really
14:46:48 <mattdm> this was kind of blocked on figuring out how the fedora-release packages would work. so it wasn't really ready to get started
14:46:53 <jzb> mattdm: sorry
14:47:07 <mattdm> jzb np. see above re atomic?
14:47:22 <jzb> atomic image? I can head it up, I am not sure I will be able to do all the heavy lifting technically, but I hope to be able to herd cats
14:47:41 <mattdm> jzb cat-herding and communication is *exactly* what we need
14:47:55 <mattdm> #action jzb to take over co-ownership of Convert Fedora Cloud Image to Fedora Server ("Adopt Your Cattle")
14:48:15 <mattdm> #info I know that's out of place in the meeting logs. life is messy sometimes. :)
14:48:26 * sgallagh perks up
14:48:49 <mattdm> #info on adopt-your-cattle: contingency plan calls for deadline at beta freeze. I think we can hit that.
14:49:15 <mattdm> #info if not, really could be added as a new package even _after_ release. we just miss out on docs and publicity
14:49:32 <sgallagh> mattdm: Do we want to discuss it here? I have some ideas.
14:49:52 <mattdm> sgallagh nope. :) just going through the list of changes and making sure they have owners and are good shape
14:50:00 <mattdm> let's _do_ discuss it soon though
14:50:15 <sgallagh> Ack
14:50:32 <mattdm> #topic Change: Smaller Cloud Image Footprint
14:50:38 <mattdm> #link https://fedoraproject.org/wiki/Changes/Smaller_Cloud_Image_Footprint
14:50:55 <mattdm> #info this subsumes a number of items from the https://fedoraproject.org/wiki/Cloud_Changelist
14:51:23 <mattdm> #info some parts of this are done -- kernel-core is awesome, cloud-init dep chain sucks less
14:51:40 <mattdm> #help smaller cloud image footprint needs a new owner, though
14:52:04 <mattdm> this is another one sandro had. it is kind of a grab-bag.....
14:52:18 <mattdm> #info the big things left are docs and i18n
14:53:13 <mattdm> #action mattdm to ask about owner for smaller cloud footprint on list
14:53:42 <mattdm> #topic Change: Big Data Cloud Image
14:54:03 <mattdm> number80: ?
14:54:56 * mattdm may be talking to myself?
14:55:07 * roshi is still here
14:55:17 <roshi> just hasn't had anything relevant :)
14:55:22 <mattdm> :)
14:55:30 <jzb> mattdm: you want to put me down for that at least interim?
14:55:43 <jzb> mattdm: it's basically we need to identify ways to slim the package set down, yes?
14:55:45 <number80> sorry
14:56:03 <mattdm> jzb the previous one, yeah? :) sure.
14:56:05 <number80> well, we're seeing with the big data sig which packages to inclkude
14:56:20 <jzb> mattdm: yeah, sorry
14:56:22 <mattdm> #action (again on previous -- jzb to look at the image-slimdown grabbag)
14:56:26 <number80> I haven't heard from jeid since and no big data sig meetings for a while
14:56:29 <mattdm> jzb np
14:57:35 <mattdm> number80: hmmm. can you restart that?
14:58:17 <number80> mattdm: yup
14:58:35 <mattdm> #info this is in stalled state
14:58:38 <number80> I'll email the big-data & cloud list about it
14:58:52 <mattdm> #action number80 to restart conversation with big data sig
14:58:59 <mattdm> it's time to get non-theoretical :)
14:59:12 <number80> *nods*
14:59:30 <mattdm> #topic Change: Docker Container Image
14:59:40 <mattdm> there was some discussion about this recently
14:59:47 <mattdm> #info we are cleared by legal to go ahead
15:00:04 <imcleod> mattdm: As in, pushing something to the official docker.io registry?
15:00:08 <mattdm> #action mattdm to ping lsm5 and dgilmore for feature update
15:00:13 <mattdm> imcleod: yes
15:00:37 <mattdm> and getting our beautiful logo on https://registry.hub.docker.com/
15:01:30 <mattdm> #info there may be some koji patches required
15:02:06 <mattdm> as I understand it, jgreguske *has* those patches made and they may already be used internally... just need to get them into fedora's koji and of course actually doing something
15:02:41 <imcleod> mattdm: That's about right, though there is the much deeper question of how to create the smallest most docker-like image possible.
15:03:08 <mattdm> imcleod: yes. for now, priority is on *getting something*. beauty can follow. :)
15:03:15 <imcleod> mattdm: +1
15:03:45 <mattdm> #topic Anything Else?
15:04:07 <jzb> mattdm: just everybody here be on the lookout for an email about new meeting times.
15:04:16 * roshi has nothing
15:04:17 <jzb> mattdm: also, who's going to Flock?
15:04:18 <mattdm> #info be on the lookout for an email about new meeting times
15:04:28 <jzb> we should plan a BoF or something there.
15:04:36 <jzb> EOF
15:04:55 <mattdm> #info we should definitely have a cloud sig get-together at flock
15:04:57 <imcleod> jzb: I just booked flights and a dorm.
15:05:08 <number80> I'll be there too
15:05:22 <mattdm> me too, ftr :)
15:06:01 <mattdm> and look! only 5 minutes over the hour
15:06:06 <mattdm> which is not bad considering the slow start :)
15:06:09 <mattdm> thanks everyone!
15:06:16 <number80> thanks !
15:06:24 <roshi> :)
15:06:30 <mattdm> #endmeeting