21:00:27 <rbergeron> #startmeeting Cloud SIG
21:00:27 <zodbot> Meeting started Thu Mar  3 21:00:27 2011 UTC.  The chair is rbergeron. Information about MeetBot at http://wiki.debian.org/MeetBot.
21:00:27 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
21:00:31 <rbergeron> #meetingname Cloud SIG
21:00:31 <zodbot> The meeting name has been set to 'cloud_sig'
21:00:34 * rbergeron goes to gather.
21:00:37 <rbergeron> #topic Roll Call!
21:00:46 * mmorsi says hi
21:00:54 * jlaska lurks
21:01:21 <rbergeron> jlaska :)
21:01:24 * jsmith-busy lurks as well
21:01:32 <rbergeron> brianlamere :)
21:01:33 <brianlamere> Yo!
21:01:50 <brianlamere> and here's to hoping Robin got the email I sent too?
21:02:02 * clalance here
21:02:32 <rbergeron> hey clalance, and jdarcy :)
21:02:47 <rbergeron> brianlamere: yes, robyn got it. Now, geting me to translate that into something ;)
21:02:47 <jdarcy> Howdy.
21:02:56 <clalance> rbergeron: Hello :).
21:03:00 * rbergeron wonders if jforbes is about, or still sick
21:03:52 <brianlamere> oops!  (bows to owner of the correctly-spelled Robyn)
21:04:14 <rbergeron> ;D
21:04:16 <rbergeron> no worries.
21:04:19 <rbergeron> It happens.
21:04:34 <rbergeron> Well. It looks like we haz a nice crowd here, so I'll get started.
21:04:44 <rbergeron> #topic EC2
21:05:11 <rbergeron> Well: Justin's not here, but I'll poll everyone anyhow on the mail that came through on the list this week about the Spins Stuff, to see if everyone's up to speed.
21:06:58 <rbergeron> I'll take that as a "everyone read that mail" ?
21:07:01 <rbergeron> Hey obino.
21:07:05 <clalance> Si.
21:07:06 <obino> hi there
21:07:34 <rbergeron> So I guess the bottom line here is: There's some question as to whether or not we're "officially" a spin, or if there are other, as-of-yet unmade rules for us to follow.
21:07:42 <rbergeron> Either way, we should probably be working a little more to do testing, etc.
21:08:02 <rbergeron> In the meantime: jforbes wrote back to a few of those mails, so we'll be keeping an eye on it.
21:08:41 <rbergeron> jlaska is presumably here, because in addition to being awesome, he's someone who could help us out (if he has time) with coming up with some of those testing things around the time that we'd be uploading images to ec2, etc.
21:09:27 <rbergeron> Questions, comments, flames?
21:09:31 <jlaska> hi!  I spoke to jsmith earlier this week who gave some awesome criteria for what can be considered a "good" ec2 image
21:10:00 <rbergeron> Is this criteria listed perhaps in an email for others to read? Or on a wiki page for folks to review? :D
21:10:01 <jlaska> I'd like to start documenting, and understanding, them a bit more ... are there links/docs I can absorb to come up to speed?
21:10:03 * rbergeron doesn't mean to flog of course
21:10:05 <jforbes> Sorry I am a bit late
21:10:06 <rbergeron> Ah.
21:10:11 <jlaska> rbergeron: not yet ... https://fedoraproject.org/wiki/User:Jlaska/Draft3
21:10:11 <rbergeron> Hey justin.
21:10:17 <rbergeron> jlaska: you da man.
21:10:27 <rbergeron> jlaska: Let me pull up some linky magic for ya.
21:10:28 <jlaska> rbergeron: well, jsmith is ... I'm just wikifying it
21:10:41 <jlaska> I need more data, and then I want to create test cases and a matrix out of that
21:10:46 <jlaska> (at least, that's the initial thought)
21:11:10 <rbergeron> #info Jlaska and jsmith have been brainstorming on ec2 criteria.
21:11:50 <rbergeron> jlaska: here's a few links.
21:11:53 <rbergeron> http://fedoraproject.org/wiki/Publishing_image_to_EC2
21:12:20 <rbergeron> http://fedoraproject.org/wiki/Cloud_SIG <-- has some info as well.
21:12:34 <rbergeron> https://fedoraproject.org/wiki/Cloud_SIG/EC2_Images <-- may be helpful for you as well.
21:12:44 <rbergeron> jforbes: What are your thoughts on the email earlier this week?
21:13:05 <jforbes> rbergeron: I think I agree with most others, we are not a spin
21:13:33 <jlaska> rbergeron: thanks
21:13:33 <jdarcy> Dare I raise the spectre of EC2 images for spins?
21:13:50 * rbergeron nods
21:13:53 <jforbes> rbergeron: but this was started because that is where koji can drag images.  We do need a formal QA process, and perhaps we can work together there
21:13:55 <rbergeron> jdarcy: I don't know. DO YOU? :)
21:14:18 <rbergeron> jlaska: So tell me: What would next steps be?
21:14:20 <jdarcy> I'm not that brave.
21:14:37 <gholms> I demand context!
21:15:08 <rbergeron> jforbes: Is there some particular reason why we didn't have selinux enabled?
21:15:12 <jforbes> rbergeron: in the meantime, we do have kickstarts in git://git.fedorahosted.org/git/cloud-kickstarts.git
21:15:16 <jforbes> rbergeron: we do
21:15:25 <jforbes> rbergeron: we didn't because we used to not include @core
21:15:33 <rbergeron> Ah
21:15:36 <jforbes> but right now it is permissive
21:15:40 <rbergeron> Okay.
21:16:36 <jlaska> rbergeron: well, it's easy to start defining tests, but that can get out of control without a goal in mind.  My initial thought would be to agree on what a "good" cloud fedora image looks like.  Assuming that's the area ya'll are looking to focus on?
21:16:37 <rbergeron> jlaska: I think the only thing I'd consider adding to what you have is maybe that we need to think a bit about "how to test" - ie: what do you need
21:17:04 * rbergeron looks around at everyone else, who is far smarter than her, for answers to jlaska's question
21:17:06 <jlaska> rbergeron: yeah, I don't know how to do that yet ... I'll need to get my hands dirty, or have someone walk me through what is needed
21:17:13 <gholms> Is it permissive for debugging purposes, or...?
21:17:55 <jforbes> gholms: yes
21:18:02 <rbergeron> brianlamere, anyone else have thoughts?
21:18:37 <gholms> I think the testing plan should include testing with selinux on as well.
21:18:41 <brianlamere> lookin back, sec
21:18:44 <jforbes> gholms: it should be enforcing before final release, but given limited EC2 debuging info it is much better to look at logs from permissive than just have things fail
21:18:49 <brianlamere> had someone come up and start talking to me
21:19:19 <jforbes> gholms: that is basially what permissive is, we get a message if it would have been blocked, instead of just outright blocking
21:19:26 <gholms> jforbes: Once booting works with it on it might be good to enable it on testing images, though.
21:19:46 <gholms> Definitely better to be permissive than disabled.  :)
21:19:51 <jlaska> how frequently are images built for ec2?  Alpha, Beta, Final ... or nightly?
21:19:59 <jforbes> gholms: 6 of 1, half a dozen of the other, there is no difference really
21:20:07 <brianlamere> I have to admit, my little cheap script doesn't start with selinux on either.  though the next version will
21:20:15 <jforbes> jlaska: alpha, beta, final... nightly would cost a fortune
21:20:46 <rbergeron> #chair jforbes gholms jlaska
21:20:46 <zodbot> Current chairs: gholms jforbes jlaska rbergeron
21:20:50 <rbergeron> #chair brianlamere
21:20:50 <zodbot> Current chairs: brianlamere gholms jforbes jlaska rbergeron
21:21:34 <gholms> Better to have it permissive so we at least get debugging info.  Thanks for switching that on.
21:22:33 <rbergeron> Do we want to talk about it more on list? Or digest for a week, and perhaps Mr. Laska would like to join us again after we've had time to eyeball it a bit?
21:23:33 <rbergeron> Okay. I suggest we do that. :)
21:23:35 <gholms> I would like to see a posting on the list.
21:23:39 * rbergeron wonders if she's lagged.
21:23:42 <jlaska> works for me :)
21:23:47 <gholms> Then I would have a better idea of what's going on here.
21:23:52 <rbergeron> jlaska: would you mind shipping a mail to the cloud list?
21:24:01 <rbergeron> Or perhaps Joining, because I know you love more mailing lists? :)
21:24:03 <gholms> rbergeron: Anything #info-able or #action-able for this topic?
21:24:07 <jlaska> #info SELinux in permissive mode for Alpha + Beta, enforcing for Final
21:24:16 <dgilmore> jforbes: quick question
21:24:22 <rbergeron> gholms: You're a chair. Feel free. ;)
21:24:22 <jlaska> rbergeron: I do love more mail!
21:24:23 <jforbes> dgilmore: yes?
21:24:31 <dgilmore> jforbes: could we make say a 2gb image.
21:24:37 <gholms> rbergeron: I can't when I jumped in near the end of the conversation.  :|
21:24:39 <dgilmore> and have the user resize it to be bigger
21:24:44 <jforbes> dgilmore: not really
21:24:59 <rbergeron> #action jlaska join cloud list and send over link, etc. for more discussion.
21:25:02 <rbergeron> https://admin.fedoraproject.org/mailman/listinfo/cloud
21:25:05 <gholms> The whole point of an EC2 image is to do your persistent stuff on a EBS volume.
21:25:05 <brianlamere> resizing is pretty easy, why wouldn't that be ok?
21:25:07 <dgilmore> jforbes: ok.  seems wasteful to upload 10gb of mostly nothing
21:25:07 <jforbes> dgilmore: I mean we could (and did) and people complained that it was not enough
21:25:28 <jforbes> dgilmore: amazon seems to compress it when they bundle the images
21:25:33 <dgilmore> jforbes: or maybe we make sparse images
21:25:44 <brianlamere> hmm...you're right.
21:26:03 <gholms> Won't it still upload the entire thing non-sparsely?
21:26:22 <jforbes> gholms: I haven't looked, though upload times don't seem to be on par for a 10g image
21:26:52 <brianlamere> but remember, during instance creation you can specify anything that is at least as large as the base.  a little block-device-mapping sda=20, etc viola you have a 20g drive
21:27:12 <clalance> jforbes: Yeah, you are right.
21:27:17 <gholms> I'm not convinced the root filesystem's size matters since it's throwaway anyway.
21:27:26 <brianlamere> if it uploads it compacted then yes, it doesn't matter.
21:27:29 <clalance> jforbes: I have a F-13 image that I built myself (basically minimal), and it is 25 parts * 10MB.
21:28:27 <gholms> [time passes]
21:29:10 <clalance> rbergeron: I think we are up for the next topic.
21:29:14 <obino> gholms: it matters only if it's too small and you cannot install packages
21:29:20 * gholms nods
21:29:24 <rbergeron> Okay.
21:29:28 <gholms> 2G ought to be fine, IMO
21:29:31 <gholms> What's next?
21:29:40 <brianlamere> maybe 5, to compromise
21:29:56 <rbergeron> #topic Aeolus
21:29:56 <gholms> As long as it's shrinkable.  :)
21:30:04 <rbergeron> I'll pick clalance, since he's here prodding me along. ;)
21:30:11 <gholms> So *that's* how you spell it!
21:30:14 <brianlamere> can't shrink, can only go up (er...once it's an image, at least)
21:30:17 <rbergeron> clalance, mmorsi: How's it going? :)
21:30:20 <clalance> rbergeron: Haha.
21:30:25 <rbergeron> Any exciting news to share? :)
21:30:34 <mmorsi> ah pretty good
21:30:38 <obino> gholms:   Aeolus
21:30:38 <obino> n 1: god of the winds in ancient mythology
21:30:41 * clalance lets mmorsi answer
21:30:50 <rbergeron> I'm already harassing people about F16. :)
21:30:53 <mmorsi> we're just wrapping up another dev sprint
21:30:54 <gholms> obino: Which ancient mythology?
21:31:02 <obino> gholms: lol
21:31:03 <clalance> gholms: Greek.
21:31:09 * gholms looks it up
21:31:20 <obino> gholms: you know where I come from, it can only be greek or roman :)
21:31:20 <clalance> gholms: It's the god that gave Odysseus the winds in a bag so he could sail home safely.
21:31:36 <clalance> s/winds/4 winds/
21:31:39 <gholms> Oh, that guy!
21:31:49 <rbergeron> openwindinabagstackops.com ?
21:32:06 <obino> it seems not taken
21:32:13 <gholms> mmorsi: What are you working on?
21:32:15 <mmorsi> some new features include:
21:32:15 <mmorsi> * automatic ssh key uploading on instances started by aeolus
21:32:15 <mmorsi> * tighter conductor / condor / image warhouse integration, they are more context aware and fault tolerant
21:32:15 <mmorsi> * realm 2 realm mapping and management in conductor and scheduler
21:32:15 <mmorsi> * finished rename (all tooling and components have been renamed)
21:32:15 <mmorsi> * more private cloud support / start of cross-cloud support
21:32:50 <mmorsi> our next sprint will be a cleanup / organizational sprint during which we'll cut another release
21:32:51 <jdarcy> I thought it was become someone said "you guys are such aeolus"
21:32:57 <obino> mmorsi: you mentioned private cloud support: is thre anything that Eucalyptus can do to help?
21:33:02 <clalance> jdarcy: Equally likely ;).
21:33:05 <gholms> #info Aeolus team is working on a metric ton of new features
21:33:11 <spot> jdarcy: hey now, watch your language.
21:33:42 <rbergeron> mmorsi: So you know at some point I have to ask about some of the parts of aeolus that are not yet in Fedora :)
21:33:54 <gholms> What parts *are* in Fedora?
21:34:03 <clalance> gholms: A number of the rubygem dependencies.
21:34:06 <clalance> But not much else.
21:34:08 <mmorsi> obino: hrm not sure, would shoot an email to aeolus-devel, we have drivers and what not, but functionality can always be expanded / improved
21:34:26 <mmorsi> rbergeron there are review requests for the deltacloud core components
21:34:32 <mmorsi> i believe the others are still a little ways out
21:34:32 <obino> mmorsi: kk will do
21:34:33 <brianlamere> how does something like CloudFormation fit in with Aeolus, btw?  Is the project going to look at it to maybe speed things along?
21:34:46 <rbergeron> mmorsi: ORLY
21:35:00 <rbergeron> mmorsi: Perhaps a mail to the list with some of those BZ's might help things along a bit.
21:35:04 <mmorsi> well ya due to deps and what not
21:35:06 <clalance> obino: There were patches posted recently to support Euca in deltacloud.
21:35:07 <mmorsi> sure
21:35:08 <rbergeron> Right.
21:35:19 <clalance> obino: Once those are in, it should be fairly easy to get Eucalyptus working.
21:35:27 * rbergeron lends spot her whip for cracking down on things
21:35:33 <clalance> (though there will probably be slight differences that will have to be taken into account)
21:35:35 <obino> clalance: cool
21:35:48 <rbergeron> clalance: That's interesting.
21:35:53 <rbergeron> That seems like it might be a fun thing to test out.
21:35:59 <rbergeron> Two cloud products, WORKING TOGETHER
21:36:02 <rbergeron> zomg.
21:36:02 <clalance> rbergeron: Haha.
21:36:06 <obino> lol
21:36:10 <mmorsi> brianlamere: i'm not terribly familiar w/ CloudFormation myself srry
21:36:10 <gholms> CloudCloud
21:36:25 <rbergeron> Eucaeoluslyptus?
21:36:30 <clalance> brianlamere: Yeah, I saw that CloudFormation stuff.  We've looked at it briefly.
21:36:38 <spot> rbergeron: i think my son has that dinosaur.
21:36:43 <rbergeron> RAWR!
21:36:44 <clalance> brianlamere: We have some stuff in the works that is similar, but a bit more generic, than CloudFormation.
21:36:47 <mmorsi> brianlamere: ya can see how that could be fit in though
21:36:48 <rbergeron> What's CloudFOrmation?
21:36:51 * rbergeron goes to find a link
21:37:06 <rbergeron> Oh, More amazon fun.
21:37:06 <gholms> rbergeron: Script your cloud creation
21:37:10 <clalance> But we could probably use CloudFormation to implement our EC2 backend.
21:37:12 <clalance> It is still unclear.
21:37:16 <brianlamere> RobYn:  you create a template that defines your cloud, then you deploy it all at once
21:37:50 <gholms> clalance: If you use that then will that break eucalyptus compatibility?
21:37:52 <brianlamere> I have things that are a bit too involved to really do in CloudFormation, but if I used it I could skip several steps
21:38:10 <rbergeron> #link http://news.cnet.com/8301-1001_3-20036390-92.html
21:38:14 <clalance> gholms: Not really, no.  One of our goals is not to be locked to any particular cloud.
21:38:26 <clalance> gholms: So we have other things planned for non-EC2 clouds anyway.
21:38:33 <clalance> (I can go into morbid detail here, if you want)
21:38:47 <gholms> clalance: I would just hate for you guys to have to maintain two completely separate code paths for AWS and Eucalyptus when they don't have to differ that much.
21:38:55 <clalance> gholms: Yeah, understood.
21:39:50 <rbergeron> Okay.
21:40:12 <rbergeron> #action mmorsi send a mail of deltacloud BZ's to the list that are looking for review love.
21:40:17 <rbergeron> Anyting else?
21:40:25 <mmorsi> thats all from here
21:40:30 * rbergeron continues to appreciate the presence of the aeolus folks at these meetings. :)
21:40:39 <clalance> rbergeron: Yeah.  Sorry we don't have a lot more fedora-specific to say.
21:40:44 <mmorsi> np good to attend
21:40:46 <clalance> The good news is that things are really starting to work.
21:41:35 <rbergeron> clalance: it's all good. I know you guys will get there eventually, and in the meantime, there are folks here who are willing to hear all about it and maybe even pitch in or give some feedback along the way.
21:42:08 <rbergeron> #topic CloudFS
21:42:11 <mmorsi> yes and we're startinjg to work towards public aeolus infrastructures from project management to public test / demo and prod servers
21:42:16 <rbergeron> Mr. Darcy :)
21:42:18 <mmorsi> thats all really :-)
21:42:28 <rbergeron> Or, i suppose i should say, mr. jdarcy, so it pings you.
21:42:28 <jdarcy> The CloudFS Road Show is finally winding down.
21:42:31 <rbergeron> Ahhh.
21:42:32 <rbergeron> Road show?
21:42:33 * gholms smacks rbergeron with a large trout
21:42:53 <jdarcy> Was at FAST and customers/partners two weeks ago, then at LF's End User Summit this week.
21:42:55 * clalance finished reading Pride and Prejudice not too long ago ;).
21:43:10 <jdarcy> Got to present on CloudFS at the latter.  Was well received.
21:43:27 <jdarcy> On the development side, Kaleb has started and gotten off to a running start improving the config/management stuff.
21:43:27 <rbergeron> jdarcy: Cool. I'm glad to hear it.
21:43:37 <rbergeron> Did you tell them they'll see it in Fedora First? :D
21:44:04 <jdarcy> Yes, I kind of did, actually.  I said something about "shooting for Fedora 16 and then some mumblemumble RHEL after that"
21:44:11 <rbergeron> :)
21:44:35 <rbergeron> Well, invite Kaleb to join the cloud sig mailing list ;)
21:44:38 <jdarcy> That's pretty much it, I think.  Should pick up more now that I'm not traveling all the time.
21:44:39 <rbergeron> Glad to hear you got another person in.
21:44:46 <rbergeron> Yeah. I hear that.
21:45:14 <rbergeron> #info CloudFS has been on a roadshow, with good feedback.
21:45:33 <rbergeron> #info Kaleb has started and Robyn can't wait to see ihm join the mailing list :D
21:45:36 <rbergeron> :D
21:45:37 <jdarcy> I'll go bug Kaleb about joining us.  He's out this afternoon, but maybe I can drag him in next week.
21:45:43 <rbergeron> #topic Any Other Business?
21:45:49 <rbergeron> I feel like I'm forgetting something. :)
21:45:59 <gholms> Open floor?
21:46:01 <brianlamere> you indicated that the ec2 calc wasn't clear
21:46:05 <gholms> Oh, this is open floor.
21:46:08 <gholms> (Sort of)
21:46:16 <rbergeron> Open floor, any other business, I consider them the same.
21:46:21 * jgreguske begins break dancing
21:46:27 * gholms needs to pay more attention
21:46:35 <rbergeron> brianlamere: I got it. It looked pretty cheap.
21:46:39 * rbergeron pops that open real uicklike
21:47:16 <brianlamere> ok - yeah, it is.  seems they beat inflation too, cause they keep making it cheaper :P
21:47:50 <rbergeron> So I guess maybe what I need to ask you to do is perhaps follow that up with an email about "what it is exactly that you're planning on doing and how that helps out"
21:48:44 <rbergeron> spot: Did I ask you Friday night about if there was a separate Fedora engineering EC2 account?
21:48:56 <rbergeron> I feel like I did, but for some reason I can't remember the answer.
21:49:33 <brianlamere> ok, I'll give a detailed plan, with what sort of access would be required
21:49:48 <rbergeron> brianlamere: cool.
21:50:01 <rbergeron> Yeah, and then we'll hash out the whole access thing.
21:50:15 * rbergeron feels like she might be lagged.
21:50:28 <rbergeron> Maybe not.
21:50:33 * gholms hands rbergeron a pair of +2 Boots of Speed
21:50:38 <rbergeron> Oooh!
21:50:48 <rbergeron> Wait, these are made of plate, and I can only wear leather and cloth
21:50:59 <rbergeron> Okay.
21:51:01 <rbergeron> Anyone else have anything?
21:51:09 <gholms> Whoever heard of plate boots of speed?
21:51:15 <gholms> Them things are leather!
21:51:20 * rbergeron will close out momentarily if not
21:51:41 <gholms> [You hear the sound of bubble wrap from upstairs]
21:51:55 <rbergeron> Funny, my house has one floor.
21:52:02 <rbergeron> Thanks for coming, guys. :)
21:52:03 <gholms> Better check the roof, then.
21:52:11 <rbergeron> Will do.
21:52:13 <rbergeron> #endmeeting