19:00:14 <rbergeron> #startmeeting Cloud SIG
19:00:14 <zodbot> Meeting started Fri Jun 24 19:00:14 2011 UTC.  The chair is rbergeron. Information about MeetBot at http://wiki.debian.org/MeetBot.
19:00:14 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
19:00:19 <rbergeron> #meetingname Cloud SIG
19:00:19 <zodbot> The meeting name has been set to 'cloud_sig'
19:00:30 <rbergeron> #topic Who's here?
19:00:35 <msavy> hello
19:00:45 * clalance here
19:00:50 * mmorsi says hi
19:00:52 * jforbes is here (and ready for vacation to start)
19:00:52 * kkeithley nods
19:01:34 <rbergeron> oh look
19:01:44 <rbergeron> everyone i just beeped at in another window is already here
19:01:45 <rbergeron> ;)
19:01:50 <jforbes> hehe
19:01:52 <gholms> What's going on?
19:01:54 <rbergeron> except gholms, who i was sure was already here.
19:02:14 <rbergeron> Well. I want to make sure justin doesn't miss out on a momento of vacation, so.....
19:02:15 <gholms> Oh, cloud meeting.
19:02:16 * gholms is here
19:02:22 <rbergeron> #topic EC2
19:02:32 * gholms has to wander off very shortly; sorry
19:02:34 <rbergeron> #chair jforbes gholms ke4qqq
19:02:34 <zodbot> Current chairs: gholms jforbes ke4qqq rbergeron
19:03:00 <rbergeron> jforbes: so what's the scoop these days :)
19:03:11 <jforbes> Sooo... progress is being made on the koji side, after a good inspection of what boxgrinder is doing in it's ec2 post, I am making some changes to the kickstarts
19:03:26 <rbergeron> wahoo!
19:03:28 <jforbes> Unfortunately koji is still slow, getting dgilmore's time is fairly difficult these days
19:03:30 * ke4qqq pretends to be here.
19:04:18 <jforbes> last koji build I did took 2h 18m just to run the build (that doesnt include the download, bundle, push, publish that comes afterward).  And that was onyl running 1 arch
19:04:44 * rbergeron nods
19:04:49 <gholms> What is the benefit of building in koji?
19:05:04 <jforbes> For reference, it took less than 1 hour to publish updated S3 and EBS backed images in BG for both x86_64 and i386, serialized, not parallel builds
19:05:13 <rbergeron> gholms: we can automate it along with building all the rest of the images and such on release day.
19:05:18 <rbergeron> AIUI
19:05:19 <rbergeron> :)
19:05:21 <gholms> Oh, nice
19:05:29 <jforbes> gholms: rel-eng does the builds that way
19:05:41 <rbergeron> yeah, i won't have to harass jforbes, and he won't have to hate me :)
19:05:45 * rbergeron hugs jforbes
19:05:51 <ke4qqq> gholms: releng likes using koji to do builds - to the point they don't want to do it any other way.
19:05:58 <jforbes> Speaking of which, updated S3 and EBS backed images are live and on the wiki
19:06:00 <ke4qqq> (which is fine, they are doing the work, so they get to choose)
19:06:07 <jforbes> us-east-1    i386    instance store  ami-6ebe4507
19:06:07 <jforbes> us-east-1    x86_64  instance store  ami-72bf441b
19:06:07 <jforbes> us-east-1    i386    EBS Backed  ami-60bd4609
19:06:07 <jforbes> us-east-1    x86_64  EBS Backed  ami-2abf4443
19:06:26 <jforbes> rbergeron: more importantly, I wont have to do the builds either.
19:06:46 <rbergeron> right.
19:07:19 <rbergeron> jforbes: are those s3 and ebs images "official" or test? (or does that matter to those two categories, or only matter for ec2?)
19:07:45 <msavy> jforbes: nice stat there with regards to boxgrinder speed
19:07:53 <jforbes> rbergeron: the only reason we were calling any of those images "test" is because they were not built in koji
19:08:04 * rbergeron nods
19:08:40 <jforbes> In all honesty, I could copy them to the other availability zones, and people would never care where the images were built
19:09:30 <rbergeron> jforbes: so you'll be re-uploading those images again??
19:09:43 * rbergeron just wants to make sure they're still test (since that's what they're list as on the wiki)
19:09:48 <rbergeron> and if not, delete that magic test word
19:10:00 <jforbes> I will be uploading images for F15 when koji builds them
19:10:14 <rbergeron> okay
19:10:17 <jforbes> Or I guess we could not do that, and just retarget for F16
19:10:36 <rbergeron> jforbes: no, i think some people are looking for F15 images
19:11:00 <jforbes> Or do that, but leave these as the official, and make the koji built images "test" for the process and have koji do the official F16
19:11:24 * rbergeron looks around for thoughts clearer than her own?
19:11:54 <jforbes> rbergeron: for people looking to actually use F15 images, these are suitable for use
19:12:03 <clalance> Yeah, I think getting F-15 from koji would be nice, if possible.
19:12:04 * jsmith cheers!
19:12:19 <clalance> (but I'm not doing the work, so I'm not going to make a fuss or anything ;)
19:12:30 <rbergeron> so these are suitable, but just not from koji - so do we want to remove the "test" word from the wiki??
19:12:40 <rbergeron> i mean seeing "test" makes me sort of freak out as a consumer of something
19:12:50 <rbergeron> i feel like it means "beta" in some ways
19:13:28 <ke4qqq> uhhhh before it gets called 'official' don't you need releng to sign off on it?
19:13:34 <ke4qqq> isn't that the process?
19:13:48 <jforbes> clalance: Yes, we need to do that, to get the process down for F16, but in all honesty, I trust these images a lot more than I trust what koji will spit out anyway, since koji hasn't spun a successful image yet, and bg image creation is well tested
19:13:57 <jforbes> ke4qqq: F14 weren't
19:14:00 <clalance> LOL
19:14:03 <clalance> Love it.
19:14:29 <ke4qqq> jforbes: yeah this is since then, a process has been defined for runnable/installable things called fedora
19:15:42 * jsmith has plenty of headaches with random contributors building random Fedora images
19:15:56 <jforbes> It's all relative, in amazon's eyes, they are official just by virtue of being on the account they are
19:16:25 <rbergeron> jforbes: I think it's more about official in the eyes of fedora folks, that everyone is beholden to the same processes and getting signoffs by rel-eng, etc.
19:16:55 * ke4qqq should restate - technically things called 'fedora' need to go through design, releng, and qa, and potentially get TM blessing from the board.
19:16:56 <jsmith> jforbes: We're talking about "official" in terms of the Fedora Board, not what Amazon thinks
19:17:13 <jforbes> I would much rather do that too.  But like I said, just image creation is over 2 hours per arch, total round trip for test spins is 9 hours which is past ridiculous
19:17:29 <jforbes> and getting it profiled is not happening
19:17:31 <rbergeron> ...and if they're broken, start over again.
19:17:43 <jforbes> right
19:18:32 <rbergeron> So I suppose the question is: Do we want to take these images that were created iwth bg, get rel-eng to sign off on it (and the other groups, though qa has no way to "test" and we have no test plan, and design isn't entirely applicable), so we can remove the word test from the wiki page?
19:18:43 <rbergeron> Or.......
19:18:55 <rbergeron> The alternative is, we wait for koji to build things, and then... still have to do the same process.
19:19:34 <rbergeron> thoughts
19:19:36 <rbergeron> anyone
19:19:39 * jforbes has no opinion on that matter, since it is a policy decision, not a technical one
19:20:19 <jforbes> Either way has zero impact on the timeline of getting koji images done
19:21:10 <rbergeron> ke4qqq, jsmith, thoughts?
19:21:30 <ke4qqq> well I guess the first question is 'is releng going to balk at a non-koji build
19:22:30 <rbergeron> dgilmore: you about?
19:22:56 * rbergeron wonders the same about qa, to some degree
19:23:11 <jsmith> Sorry, on the phone... let me read the scrollback
19:23:33 <rbergeron> okay, dennis isn't here.
19:23:42 <rbergeron> I'll copy him on meeting notes.
19:23:46 <ke4qqq> spot: willing to weigh in for releng?
19:24:00 * rbergeron isn't sure spot is in today, but not positive
19:24:02 <jforbes> Yeah, he has been busy lately
19:24:11 <jforbes> he = dgilmore
19:24:14 <rbergeron> jforbes: yeah
19:24:18 <rbergeron> and you guys are both out next week.
19:24:23 <jforbes> right
19:24:28 <rbergeron> So I think we're not getting anything immediatley either way.
19:25:04 <jforbes> Well, like I said, this decision has zero impact on the timeline of getting images from koji.  It is more of a question as to whether or not they are worth waiting for
19:25:08 <rbergeron> But I'd like to get this solved, because either way, even if it's getting spit out of koji, it will still need the whole signoff.
19:25:13 <rbergeron> yeah.
19:26:10 <rbergeron> sooooo
19:26:18 <jsmith> Right...
19:26:20 * rbergeron sighs and takes a deep breath and tries to figure out what she's concluding to here
19:26:30 * gholms hands rbergeron an #action
19:26:40 <jsmith> The idea of having Koji build the images is to eliminate the human factor as much as possible
19:26:46 <rbergeron> I guess (1) is, who is going to take this stuff to qa, board, design, rel-eng
19:26:53 <jsmith> Obviously that isn't work as well as we'd like
19:27:16 <ke4qqq> jsmith: if that's all it is, bg works for that.
19:27:31 * jforbes notes bg has less human factor than koji
19:27:36 <jsmith> ke4qqq: Yes, as long as rel-eng is fine with using bg
19:27:47 <rbergeron> ke4qqq: yes, but rel-eng doesn't want to have to learn new tooling, i think is the issue.
19:27:48 <jforbes> since koji only builds the images, humans still have to do the rest
19:27:53 <jsmith> The images need to be created *by* rel-eng
19:28:17 <jforbes> bg goes from creation to publish with 1 command, you just have to make them public when it is done
19:28:53 <jforbes> The fact is, the community wont wait for koji images anyway, they are already using the bg created images, and folks who haven't found those are creating their own with bg
19:29:20 <jsmith> Again, the idea of having koji create the images would be that they'd be ready at GA
19:30:17 <rbergeron> jsmith: yes, and I think we know that. I think what we're trying to more immediately solve is "do we want to make these images official via Process, or do we want to wait for koji imaes now and make those official, and eithe rway, who is going to drive that stuff through all the signoffs"
19:30:23 <jforbes> jsmith: right, and if we get this process down for F16, they will be (or if they ran BG to do it, the same) but we aren't asking about that, we are asking about F15, GA is long past
19:30:59 <rbergeron> jforbes and dgilmore are working through the koji issues so we don't have to have this problem continually.
19:31:17 <rbergeron> in the meantime, we have images sitting out there that aren't "official" or signed off on, but considered "test" on the wiki page.
19:31:26 <rbergeron> And i think it would be good to make *something* be official.
19:31:57 <rbergeron> okay. Soooooo:
19:32:20 <gholms> #info Problem: Fedora has not published any non-test EC2 images
19:32:28 <rbergeron> I propose that I guess I will go and take it to all the groups, starting with rel-eng about bg vs. koji for F15 stuff right now.
19:32:45 <rbergeron> after I get an answer from them, I'll hit up the other groups, and then... what, i go to the board? Or is it just done?
19:32:50 <rbergeron> I guess I'll go read the documentation on the process.
19:33:43 <rbergeron> jsmith: if you could follow up with dgilmore to see if he has responded or just talk with him about it anyhow while y'all are in brazil that would be appreciated also, though i'll send him a mail and copy the cloud list.
19:34:13 <rbergeron> But i don't want to hang on that for a whole week, really, would prefer to move forward with other stuff too. :)
19:34:19 <rbergeron> Is everyone cool with that?
19:34:47 * rbergeron cues up a gholms tumbleweed-ism
19:34:50 <jforbes> That works, the images are published, since people are using them anyway, I will copy the S3 backed to the other availability zones
19:35:15 * gholms hands rbergeron an #action :)
19:35:25 <jforbes> At that point, if a decision is made while I am on vacation, you can just change the wiki and submit the form for Amazon that makes them "published official images"
19:35:38 <rbergeron> jforbes: all the magic in EBS and ec2 and what not is all the same image, right?
19:35:41 <jforbes> (read publicises them on their main ec2 mpage_
19:35:41 <rbergeron> no differences?
19:36:01 <rbergeron> jforbes: is there anything that amazon does to bless them as official on their end?
19:36:04 <jforbes> rbergeron: yes and no... Same images, but bound to a different kernel
19:36:18 <jforbes> rbergeron: nope, we just tell them that it is official and they publish them.
19:36:44 <jforbes> In fact, I will have my laptop, can do that piece if I get an email.  I will be checking in the evenings
19:36:54 <rbergeron> ke4qqq, jsmith: will different kernels mean that i need two separate things to get signed off?
19:37:07 <jsmith> rbergeron: Absolutely :-/
19:37:39 <rbergeron> awesome.
19:38:03 <rbergeron> jforbes: can you shoot the list a mail with .. like... where I can find the two different... whatevers... that I can point qa at? on our local infra?
19:38:09 * rbergeron clearly is the technical person in the room
19:38:20 <gholms> Heh
19:38:48 <jforbes> rbergeron: where the images are? https://fedoraproject.org/wiki/Cloud_SIG/EC2_Images#Fedora_15_Test_Images lists them
19:39:08 <jforbes> rbergeron: accounts for testing, that would go through spevack
19:39:09 <rbergeron> #action rbergeron to take Amazon stuff and turn into two separate fedora "blessing" proposals to drive through process, starting with rel-eng
19:39:37 <rbergeron> #action rbergeron to talk to dgilmore about just using the BG images that are already up and making those official, vs. waiting for koji stuff to sort out
19:40:08 <rbergeron> jforbes: I guess I mean - is there a way for them to just test the images locally rather than having an amazon account
19:40:11 <jforbes> There is no way to actually fully test the images on local infra since a good bit of the functionality (like being able to log in) requires getting an ssh key from amazon on boot
19:40:59 <rbergeron> ah.
19:40:59 <gholms> Technically there is; it just isn't in Fedora.
19:41:18 <jforbes> to test locally on kvm I actually mount an image, modify it, and boot it, so it is fine for debug, not great for official image testing
19:41:47 <rbergeron> okay. well, i'll sort out the whole how to test thing too with jlaska and assume that spevack will say yes to what i ask. ;)
19:42:05 <rbergeron> Okay.
19:42:17 <rbergeron> I think we have that sorted out, as much as possible. Yes?
19:42:22 * rbergeron holds for 5 seconds before moving on
19:43:02 <rbergeron> #topic CloudFS
19:43:14 <rbergeron> jdarcy, kkeithley: if we haven't bored you to tears yet ;)
19:43:21 <kkeithley> jdarcy mentioned leaving early to take kids camping. I know he's been working on getting the auth creds from the SSL transport to do auth². Not sure what else he's been working on
19:43:29 <kkeithley> I'm working on packaging of cloudfs. fighting with libtool, finally grokked that if I install a lib outside the normal places then libtool adds an rpath (to the library. huh??? that's strange.) grrrr. And before anyone says "don't install libs outside the normal places" don't worry, I'm not. Not for fedora consumption anyway.
19:44:01 <rbergeron> ;)
19:44:28 <kkeithley> that's about it
19:44:42 <rbergeron> kkeithle: are you going to be submitting the packages?
19:44:49 <kkeithley> yes
19:44:49 * rbergeron notices that you aren't in the packaging group yet
19:45:12 <kkeithley> oh
19:45:13 <rbergeron> ke4qqq: might you be able to sponsor kkeithley when that time comes around?
19:45:23 <ke4qqq> yep
19:45:24 <rbergeron> unless kkeithley already has a prospective sponsor?
19:45:48 <kkeithley> nope, unless jdarcy in the group already?
19:46:04 <kkeithley> /in/is in/
19:46:05 <rbergeron> ke4qqq: is it a process that should be started now, or just when he gets the spec files ready?
19:46:08 * rbergeron looks at zodbot
19:46:35 <rbergeron> jdarcy isn't in packaging either
19:46:39 <kkeithley> I'll have spec files ready probably early next week
19:47:11 <ke4qqq> kkeithley: feel free to do unofficial reviews on aeolus packages - that will speed sponsorship - seems like I sent you links last meeting
19:47:11 <rbergeron> okay. get with ke4qqq or ping on list when you're ready, so we can get you going. :)
19:47:22 <rbergeron> or what ke4qqq says
19:47:39 * kkeithley nods
19:47:51 <kkeithley> I have the links from last week
19:48:07 <clalance> kkeithley: I'm also happy to trade some reviews with you.
19:48:12 <kkeithley> squirreled away
19:48:17 * clalance will start packaging aeolus in earnest next week
19:48:36 <kkeithley> clalance: thanks
19:48:37 <clalance> Well, the *rest* of aeolus, I should say.
19:48:43 <rbergeron> cool. :) thanks guys
19:48:52 <rbergeron> clalance: that sounds like fun. ;)
19:49:02 <clalance> Fun is a word.
19:49:06 <clalance> Though maybe not the one I would use ;).
19:49:20 <rbergeron> #action kkeithley to wrap up packaging, work with ke4qqq on sponsorship, sync with clalance on aeolus reviews
19:49:25 <clalance> (it must be done, though)
19:49:37 <rbergeron> yup.
19:49:43 <rbergeron> kkeithley: thanks!
19:49:54 <kkeithley> rbergeron: yw
19:49:56 * rbergeron notes we shoudl probably talk to jdarcy too about getting sponsored, if he's going to be doing packaging as well
19:50:01 <rbergeron> #topic Aeolus
19:50:14 * rbergeron looks at the clock
19:50:35 <clalance> rbergeron: Very briefly; working on getting a 0.3.0 release of aeolus out in the next couple of weeks.
19:50:49 <clalance> That's what I'll end up packaging for F-16, at least as a start.
19:50:55 * rbergeron nods
19:51:19 <clalance> 0.3.0 will have a completely revamped and spiffy UI.
19:51:21 * rbergeron notes that feature freeze is july 24,as a friendly FYI.
19:51:26 <clalance> Lots of new features and bugfixes.
19:51:31 <clalance> Thanks for the reminder.
19:51:49 <clalance> This week has been crazy, otherwise I would have started already.
19:51:55 * rbergeron knows you wanted one more countdown clock over your head, right? :)
19:52:00 <clalance> Haha, yeah.
19:52:26 <clalance> Other than that, not a whole lot to report.
19:52:30 <clalance> Just working away as usual.
19:52:50 <rbergeron> excellent. thanks :)
19:53:11 <rbergeron> #info clalance et al working on getting a 0.3.0 release of aeolus out, packaged in f-16.
19:53:20 <rbergeron> #topic Cloudstack
19:53:24 <rbergeron> ke4qqq: how goes
19:53:45 <ke4qqq> in progresss
19:53:49 <ke4qqq> two deps remaining
19:53:50 <ke4qqq> eof
19:54:07 <clalance> ke4qqq is much briefer than me ;)
19:54:10 <rbergeron> #info cloudstack is in progress.
19:54:35 <rbergeron> #topic Sheepdog
19:55:16 * rbergeron looks at ke4qqq again
19:55:19 <rbergeron> (sorry)
19:55:24 * ke4qqq has nothing to report here I don't tihnk
19:55:26 <ke4qqq> it's in.
19:55:54 <rbergeron> ke4qqq: ... maybe we could update the feature page to show more than 25%? or is there other stuff we're waiting on there?
19:56:11 <rbergeron> (like how to test info etc?)
19:56:13 <ke4qqq> ummmm yeah
19:56:19 <rbergeron> okay.
19:56:27 <ke4qqq> I'd like to get virtmanager/python-virtinst support
19:56:36 <ke4qqq> but yeah should be more than 25%
19:56:45 <rbergeron> #info Sheepdog is done, in, packaged. Other feature-related things could go in, though.
19:57:00 <rbergeron> #action ke4qqq to update sheepdog feature page at his convenience
19:57:08 <rbergeron> #topic Open floor
19:57:16 * rbergeron will do this in an hour, darnit
19:57:23 <gholms> Hehe
19:57:26 * rbergeron hopes she didn't miss anyone
19:57:33 <rbergeron> anyone have open-floor-ness?
19:57:49 * ke4qqq would like some openstack-swift testing
19:57:56 <ke4qqq> with the package i wrote to the list about
19:58:00 <ke4qqq> it's a big update
19:58:22 <gholms> Got a link for posterity's sake?
19:58:37 <ke4qqq> gholms: one second
19:59:19 <ke4qqq> #link http://koji.fedoraproject.org/koji/taskinfo?taskID=3153175
19:59:51 <gholms> #help Testers wanted for new openstack-swift version
19:59:54 <rbergeron> #info openstack-swift testing appreciated
20:00:04 * gholms wins \o/
20:00:52 <rbergeron> #link http://lists.fedoraproject.org/pipermail/cloud/2011-June/000647.html
20:01:02 <rbergeron> ke4qqq: thanks for taking that on, btw.
20:01:32 <ke4qqq> rbergeron: it's self serving :)
20:01:36 <gholms> [The clock begins chiming]
20:01:39 <rbergeron> ke4qqq: yeah, yeah
20:02:02 <rbergeron> ke4qqq: i wonder if you'd get better assistance in tsting if there was a link to openstack docs somewhere about how it could be tested
20:02:17 <gholms> +1
20:02:35 <rbergeron> unless by testing you just mean "does it install without explosions"
20:02:52 <rbergeron> anyway. food for thought there
20:03:16 <ke4qqq> perhaps I'll update the message in the next day or so.
20:03:55 * rbergeron takes the keys out of the bus ignition
20:03:57 <rbergeron> okay.
20:03:59 <rbergeron> ANYTHING ELSE?
20:04:05 * rbergeron hears nothing
20:04:19 <gholms> [The Easter Bunny flies through the window]
20:04:21 <rbergeron> thanks for coming, guys. glad to hear things are moving along for everyone as best as they can be. ;)
20:04:30 <rbergeron> gholms: WHERE DO YOU COME UP WITH THOSE THINGS
20:04:38 <gholms> rbergeron: Nerd magic.
20:05:10 <rbergeron> sweet.
20:05:11 <gholms> I really appreciated that EC2 image discussion; thanks, guys.
20:05:24 <rbergeron> gholms: we can do it again next week if you want ;)
20:05:39 <rbergeron> #endmeeting