21:00:06 <rbergeron> #startmeeting Cloud SIG
21:00:07 <zodbot> Meeting started Thu Aug 26 21:00:06 2010 UTC.  The chair is rbergeron. Information about MeetBot at http://wiki.debian.org/MeetBot.
21:00:07 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
21:00:11 <jforbes> heya
21:00:31 <rbergeron> hey jforbes, how goes it :)
21:00:34 <gholms|work> #topic Roll call
21:00:35 <rbergeron> #topic roll call
21:00:48 <rbergeron> #chair gholms|work jforbes
21:00:48 <zodbot> Current chairs: gholms|work jforbes rbergeron
21:00:50 <jforbes> It goes, hopefully generating the last F13 image set
21:01:03 <rbergeron> anyone else about?
21:01:16 * gholms|work tries to stay awake
21:01:25 * rbergeron grins
21:01:49 * jsmith is here
21:01:59 <jsmith> (but I'll probably have to leave early)
21:02:03 * rbergeron waves to jsmith
21:02:09 <jsmith> rbergeron: Long time, no chat
21:02:09 <rbergeron> #topic Agenda
21:02:16 <rbergeron> #link https://fedoraproject.org/wiki/Cloud_SIG#Upcoming_meeting_agenda
21:02:29 <rbergeron> More or less standard fare, though I've added some "let's think about the future" if we get that far into the meeting :)
21:02:34 <rbergeron> #topic EC2 status
21:02:37 <smooge> here
21:02:40 <jforbes> gholms|work: I know that feeling, going on 3 hours sleep, my oldest turned 7 today so he woke up early
21:02:47 <rbergeron> jforbes: take it away, sir :)
21:03:03 <jforbes> EC2 status... Building what I hope to be the last F13 images right now after getting more feedback
21:03:23 <rbergeron> feedback from.... amazon?
21:03:33 <jforbes> specifically Amazon wanted to see root login locked out
21:03:40 <rbergeron> ah
21:03:40 <jforbes> an ec2-user as default user
21:03:47 <rbergeron> can we document those little things somewhere? maybe on the wiki?
21:04:00 <jforbes> and an entry in /root/.ssh/authorized_keys with a forced command saying how to log into the instance
21:04:01 <rbergeron> or in your notes list for when we figure out how to make this consistently repeatable :)
21:04:26 <jforbes> Indeed (which is also why I typed them here, so there is a record of them somewhere
21:04:33 <rbergeron> #info jforbes is building what he hopes is the last F13 image right now, cleaning up some items after getting feedback from the folks at amazon
21:04:43 <rbergeron> #info amazon wanted root login locked out
21:05:12 <rbergeron> #info and an entry in root/.ssh/authorized_keys with a forced command saying how to log into the instance
21:05:23 <rbergeron> okay.
21:05:29 <rbergeron> So after this..... ?
21:05:39 <jforbes> #info Amazon requested ec2-user as the default user and a sudo nopassword entry for that user
21:06:07 <jforbes> After this, we make images public and call for broad testing
21:06:10 <brianlamere> sorry, have stuff going on at work
21:06:22 <gholms|work> Can we just call the user 'fedora' so we match both our live CD and Ubuntu's precedence?
21:06:24 <brianlamere> are we talking about Ben yet?  heh
21:06:24 <rbergeron> brianlamere: no worries, just making sure you knew :)
21:06:45 <rbergeron> brianlamere: not yet :)
21:07:12 <rbergeron> jforbes or anyone else: do we know exactly what we want to test - is there a list we can point users to of things they shoudl specifically check?
21:07:14 <brianlamere> what the user is called is irrelevant; bucket names are what is important, otherwise you're literally just a number
21:07:16 <jforbes> While the public is testing we can work with amazon to get the official bit set and work out the S3 mirror bit that Nathan had in mind
21:07:35 <rbergeron> brianlamere: just a number? where did they ever get that idea from :)
21:08:03 <brianlamere> jforbes:  was Ben someone you were already working with?  does the official account already have S3 comp'd?
21:08:20 <gholms|work> What are we using for EC2-specific init scripts?
21:08:21 <jforbes> brianlamere: not when amazon makes you a clicky :)
21:08:48 <brianlamere> yeah - that's what Ben was saying he could easily do if we just tell him the AMI number to do it with
21:08:53 <jforbes> gholms|work: rc.local bits still, but amazon would like those broken out... That requires a new package, and will be an F14 target
21:08:55 <brianlamere> since he's in the group that manages it
21:09:10 <gholms|work> jforbes: Have you looked at http://packages.ubuntu.com/karmic/ec2-init ?
21:09:20 <jforbes> gholms|work: I have
21:09:32 <gholms|work> Can we make use of something like that?
21:09:44 <brianlamere> look at some of the ec2-init stuff, they query a meta url (169.something.169.something) and get info about it
21:09:53 <jforbes> gholms|work: possibly
21:10:11 <gholms|work> brianlamere: Yeah, those are the URIs that their infrastructure gives VMs to configure themselves with.
21:10:26 <brianlamere> it's pretty basic - I use it to get the private IP address when something starts, so an instance can tell all it's friends where it can be found
21:10:39 <jforbes> and ubuntu uses cloud-user  instead of ec2-user. no reason to go against their wish their though, makes docs easier for new users
21:10:40 <brianlamere> err...its.  sorry.
21:11:25 <gholms|work> ec2-user makes sense if their docs say that.
21:12:38 <smooge> jforbes, all the things from amazon look like stuff from the initial kickstart? correct?
21:13:02 <smooge> oh sorry didn't scroll down (again)
21:13:02 <jforbes> smooge: yeah, with a bit of post
21:13:19 * rbergeron sends smooge a new scroller for the day
21:13:53 <rbergeron> Soooo - can i drag us back to testing for a moment?
21:14:11 * rbergeron notes that we could announce a testing day for EC2 - if we have all the stuff in place
21:14:21 <rbergeron> ie: people available online to help out, a standard list of things to test, etc.
21:14:35 <rbergeron> and all the bits for access that people would need in place.
21:14:48 <brianlamere> here's ubuntu's ec2 init stuff:  https://help.ubuntu.com/community/CloudInit
21:15:13 <brianlamere> the Amazon guy suggested it as something we should look at and use/fork/learnfrom
21:15:37 <rbergeron> jforbes: btw, EC2 has been picked for a Feature Profile to be done in the next few weeks, and I'm on the hook for writing it :)
21:16:03 * gholms|work is willing to package cloud-init if we want to do that
21:16:12 <jforbes> rbergeron: Yeah, I was also asked to see if we had anyone in RDU who is part of this SIG... anyone?
21:16:13 <gholms|work> rbergeron: Good luck
21:16:41 <rbergeron> ohhh, for the video stuff - I have no idea. gregdek was, I don't know about now - I suspect nobody
21:17:13 <brianlamere> ok, so something to keep in mind:  there should be a minor difference for ec2 instances over instances in the wild, outside of ec2; ec2 instances should have a package installed that covers there being different repos than normal
21:17:18 <jforbes> rbergeron: Yeah, they thought I was in Boston, but I am very far away from that
21:17:21 <rbergeron> smooge: do we have any infrastrucutre folks in RDU?
21:17:36 <rbergeron> yeah - I was like... ummm, he's on hte better half of the country, yo
21:17:40 <rbergeron> :)
21:17:41 <jforbes> brianlamere: why?
21:17:42 <gholms|work> brianlamere: Why's that?
21:17:45 <brianlamere> is there already a package for the minor extras that get put on an ec2 instance?
21:18:09 <smooge> fore Fedora?
21:18:13 <jforbes> brianlamere: no, there needs to be one created, scroll up :)
21:18:17 <smooge> rbergeron, for Fedora or RH in general?
21:18:22 <gholms|work> On Fedora?  Not yet.
21:18:33 <brianlamere> why:  because then the AWS repos that can only be reached inside amazon can actually be added as the primary repo targets
21:18:46 <gholms|work> brianlamere: Why not use mirrormanager?
21:18:47 <jforbes> brianlamere: umm, mirror manager takes care of that
21:18:48 <rbergeron> smooge: someone that might be able to talk a bit to Cloud stuff with Fedora on video
21:18:59 <gholms|work> Mirrormanager can prioritize repos by IP blocks.
21:19:09 <smooge> well I will be out there in late September for RHEL training
21:19:15 <brianlamere> ah...ok, I apologize, I didn't realize this
21:19:18 <gholms|work> (Speaking of that, do we have a list of EC2's IPs?)
21:19:23 <rbergeron> hrmmmmmm
21:19:24 <jforbes> gholms|work: not yet
21:19:35 <rbergeron> jforbes: i'll talk to stickster about the timing on the video stuff
21:19:48 <brianlamere> Ben said there are ACLs for easily making buckets only accessible within a region
21:19:56 <jforbes> rbergeron: I appreciate it, had it been 2 weeks ago, I was in boston :)
21:19:57 <jsmith> rbergeron: Or me -- doesn't have to block on stickster
21:20:11 <jsmith> jforbes: But the video team wasn't there :-(
21:20:22 <rbergeron> jsmith: ah - i didn't know he had you in the loop for the video stuff
21:20:29 <brianlamere> problem is, none of the tools out there reference those ACLs (boto doesn't, for instance).  he seemed to be suggesting they could set that up for us, though.
21:20:30 <rbergeron> jsmith: can you see what the timeline for that is?
21:20:39 <jsmith> rbergeron: It technically falls on my to-do list -- he's just helping me learn the ropes
21:20:48 <gholms|work> brianlamere: Does the web UI have that?
21:20:55 <rbergeron> I know it was probably colby hoke doing the video, esp if they were thinking RDU - but smooge is going to be there in late september and might be able to do it
21:21:03 <rbergeron> we don't have anyone else in RDU, really
21:21:19 <rbergeron> I don't know about boston - but jforbes just said they wer asking about RDU - id ont' even think any of us are in boston, even
21:21:40 <jforbes> jsmith: yeah, the good news is, this is a real community effort, the bad news is, that means no one is close to RDU or BOS
21:21:49 <jsmith> jforbes: That's OK -- we'll improvise
21:22:04 <jforbes> rbergeron: Yeah, stickster said they would be in Boston in a couple of weeks or something, but when I said I wasn't there he asked about RDU folks
21:22:04 <gholms|work> RDU?
21:22:06 <jsmith> jforbes: Worse comes to worse, we fly somebody to RDU
21:22:09 <rbergeron> send colby up to OLF :)
21:22:16 <jsmith> rbergeron: That may work as well
21:22:23 <jsmith> gholms|work: RDU = Raleigh, NC
21:22:24 <rbergeron> not that I could talk about it
21:22:25 <gholms|work> Oh, there.
21:22:29 <smooge> jsmith, when did you need it?
21:22:32 <jforbes> dhuff is one of th ose places, but he has been kind of out of the loops lately
21:22:32 <jsmith> gholms|work: (silly airport codes)
21:22:39 * rbergeron nods
21:22:49 <jsmith> smooge: End of September at the latest
21:22:52 <rbergeron> yeah, huff is in RDU - he might be able to wing it though - he was certainly involved for a while
21:23:03 <jsmith> Maybe we can get huff and smooge
21:23:08 <smooge> well I will be there 26th to the 1st
21:23:13 <rbergeron> of sept?
21:23:15 <jsmith> They could probably make something work
21:23:31 <rbergeron> jsmith: can you check it out? :)
21:23:41 * rbergeron hopes the rdu video crew of awesome will be around then
21:23:43 <jsmith> rbergeron: It's on my to-do list :-)
21:23:44 <smooge> of sept. I can bring my brightest hawaiian shirt or the one with blue/red cross hatches..
21:23:56 <rbergeron> smooge: can you layer them?
21:24:20 <rbergeron> anyway: EC2 - any more thoughts here folks?
21:24:27 <gholms|work> On the image-building front:  new euca2ools is on its way to updates-testing.  Please try building images and managing instances with it so we can find out if it works for it.
21:24:36 <gholms|work> s/it\./that./
21:25:07 <gholms|work> They managed to make it compatible with deps that are already in stable Fedora, so it's on its way out the door.
21:25:13 <rbergeron> gholms: awesome.
21:25:30 <jforbes> gholms|work: that makes me happy :)
21:25:50 <gholms|work> I also managed to get in touch with boto's author, who seemed interested in helping with packaging.
21:25:53 <gholms|work> rsc, brianlamere: ^
21:26:10 <rbergeron> gholms|work: is he on the mailing list?
21:26:12 <jsmith> gholms|work: Is that the same one that's in your personal repo?
21:26:23 <gholms|work> jsmith: euca2ools-1.3-1, yeah.
21:26:39 <jsmith> gholms|work: Cool... already have it, but haven't played w/ it yet
21:26:53 <brianlamere> gholms: cool - well, he trumps me by far in usefulness ;)  though I'd still like to help out with that
21:27:00 <rsc> gholms|work: the main question still remains: How compatible is 1.9 vs 2.0 beta?
21:27:20 <brianlamere> I did catch up with rsc finally, he's going to send me his rpms to test with.
21:27:37 <gholms|work> rsc: Good question.  I don't require 2.0 any more for my package, so I can't really tell for sure.
21:27:48 <brianlamere> oh, you're here - yeah, just send me the rpms you have and I can start testing the backwards compat starting...tomorrowish
21:29:28 <rbergeron> anything else on ec2? :)
21:29:36 * rbergeron is just here with the cattle prod, don't mind me :)
21:29:52 * gholms|work raises hand
21:30:02 * rbergeron laughs... go ahead
21:30:03 <rbergeron> :)
21:30:18 <gholms|work> I'm also contemplating building euca2ools for el5, but I need a packager for m2crypto.  Any takers?
21:30:29 <brianlamere> do we want to get moving on actually having repos put up on S3, then?  and then work with getting mirrormanager up on ec2 heads once we get that worked out?
21:30:38 <smooge> oh god.. I am trying to remember why that name calls headaches
21:30:39 <gholms|work> #info euca2ools 1.3 on its way to updates-testing; please test
21:31:25 <rbergeron> gholms|work: I can try and hunt one down if we have no takers.
21:31:43 <gholms|work> brianlamere: That sounds like a good idea.  It would be nice to have a process nailed down for getting stuff from the master mirror to S3 buckets.
21:32:20 <gholms|work> rbergeron: I just need someone to do it for el5 since el6 and Fedora have new enough Python to support euca2ools on their own.
21:32:54 <rbergeron> #action rbergeron to try and hunt down a packager for m2crypto for el5
21:33:10 <brianlamere> gholms:  yes, I told Matt I'd work with him to get MirrorMaster to understand S3 as a backend
21:33:31 <brianlamere> currently, MirrorMaster only understands a flat filesystem
21:34:01 <brianlamere> point is, there's some dev work that needs to take place to facilitate it
21:34:09 <gholms|work> There's a bit of controversy about what that package would be called, but it's basically the same thing as the existing epel-5 package, just built for python 2.6
21:34:49 <rbergeron> gholms|work: noted
21:35:10 <brianlamere> ok - m2crypto:  I sortof use it, in that I use paramiko sometimes (well, I use fabric.api, which uses paramiko; I rarely use paramiko directly).  It's really really indirectly that i use m2crypto, in other words
21:36:13 <gholms|work> (specifically it may be called m2crypto26 or python-m2crypto26)
21:36:47 <rbergeron> #info There's a bit of controversy about what that package would be called, but it's basically the same thing as the existing epel-5 package, just built for python 2.6; specifically it may be called m2crypto26 or python-m2crypto26
21:37:29 <gholms|work> Do we want to attempt to package cloud-init now, or do we need to do dev work on it first?
21:38:02 * rbergeron doesn't know - jforbes? ^^
21:39:21 <rbergeron> gholms|work: want to ask on the list?
21:39:37 <jforbes> gholms|work: It probably needs some looking at, anyone want to start the packaging and I will be glad to review
21:39:38 * gholms|work shrugs
21:39:52 <gholms|work> I'll see what I can do as far as packaging it goes.
21:40:07 * rbergeron grins
21:40:28 <rbergeron> #action gholms|work to look at packaging cloud-init, jforbes can look at / review
21:40:39 <gholms|work> I see what appears to be upstart-specific stuff, so we need to make sure that it works with systemd.
21:41:01 <rbergeron> alllrighty - anything else here?
21:41:34 <gholms|work> [You hear the sound of distant drums]
21:41:44 * rbergeron grins
21:41:52 <rbergeron> #topic openstack packaging
21:41:56 <rbergeron> silassewell, still about?
21:42:05 <silassewell> yup
21:42:49 <silassewell> For the swift stuff I'm still waiting on the reviewer
21:42:57 <rbergeron> okay - I'll get after ian.
21:43:03 <rbergeron> just the questions you had, or anything else?
21:43:14 <rbergeron> https://bugzilla.redhat.com/show_bug.cgi?id=617632
21:43:27 <silassewell> If they're still an issue, otherwise just waiting for a formal review
21:44:06 <rbergeron> #action rbergeron to follow up with ianweller on swift package review
21:44:12 <gholms|work> #info openstack-swift waiting on package review
21:44:16 <gholms|work> Gah, too slow!
21:44:20 <rbergeron> i know he's just started school, so he's a little hectic atm i'd guess. :)
21:44:41 <gholms|work> If anyone else wants to steal the review I'm sure he wouldn't mind that much.  :P
21:44:55 <rbergeron> no, i'm sure he'd be kosher with that
21:45:04 <silassewell> But reviewing packages it so extremely fun :p
21:45:06 <rbergeron> though we'd have to find a taker :)
21:45:15 <gholms|work> Ask on the list.
21:45:28 * rbergeron will do just that
21:45:38 <ianweller> hey look, a highlight
21:45:43 <gholms|work> :D
21:45:43 <rbergeron> #action rbergeron to look on list for another reviewer if ianweller is swamped :)
21:45:47 <rbergeron> ears burning? :)
21:45:54 <ianweller> yeah school has blown me up, but this weekend should be... freeish
21:46:08 <ianweller> free enough to finish up that review. i think it's mostly all been taken care of.
21:46:22 <rbergeron> ianweller: thanks :)
21:46:38 <rbergeron> #action ianweller to finish up review this weekend, time permitting :)
21:47:12 <rbergeron> alrighty.
21:47:18 <rbergeron> silassewell: thanks for all your work on this :)
21:47:42 <rbergeron> #topic Cloud in F15 and beyond......
21:48:08 <rbergeron> Sooo - I realize that we're still in F14 - but I'd like for people to start thinking about the "beyond getting on EC2" thing for F15.
21:48:32 <rbergeron> If there are things we want to do, have, and what that will take between now and then - because really, we don't have all of F15, just the time we have to get things into feature list and so forth.
21:48:39 <rbergeron> Does anyone have any thoughts / ideas / preferences?
21:48:42 <gholms|work> How about we make refinements based on what we learn with F14?
21:48:49 * rbergeron notes she would really like to see deltacloud get packaged.
21:48:54 * rbergeron nods - excellent.
21:49:06 <rbergeron> #info Refinements based on what we've learned with F14.
21:49:12 * rbergeron wonders if the idea tag works?
21:49:13 <gholms|work> Nu-uh
21:49:16 <gholms|work> #undo
21:49:16 <zodbot> Removing item from minutes: <MeetBot.items.Info object at 0x18355290>
21:49:17 <gholms|work> #idea Refine our processes based on our F14 experiences
21:49:21 <rbergeron> i think it does
21:49:23 * rbergeron grins
21:49:32 <rbergeron> #idea Get deltacloud packaged.
21:49:58 <gholms|work> #idea Document image building, publishing SOPs
21:50:20 * rbergeron notes that sdake is getting sheepdog packaged as well
21:50:44 <jforbes> nifty, saw a presentation on sheepdog at the kvm forum
21:50:52 <gholms|work> What do you think of it?
21:51:00 <rbergeron> I guess the question is, do we want to solely focus on just having images on amazon and elsewhere - or do we want to work on also having toolsets for people who are doing cloud-related development, etc?
21:51:17 <rbergeron> ie - adding in nosql databases, etc
21:51:21 <jforbes> still a lot of work to be done, but the idea is good
21:51:46 <rbergeron> #link http://www.osrg.net/sheepdog/_static/kvmforum2010.pdf
21:51:55 <rbergeron> #link http://www.osrg.net/sheepdog/
21:52:20 <jforbes> rbergeron: I think both are acceptable goals, but focus has to be on getting Fedora images into EC2 first
21:52:20 <smooge> ah thanks
21:52:26 <rbergeron> jforbes: indeed.
21:52:33 <rbergeron> smooge: ?
21:52:51 <gholms|work> rbergeron: I think people who want to do that can take one of our VMs, customize it, then build their own images out of those.
21:52:55 <smooge> link to sheep dog
21:53:00 <smooge> I was about to ask
21:53:14 <rbergeron> sheepdog has some pre-work I think to getting packaged - man pages, init scripts were missing until like 2 days ago
21:53:34 <rbergeron> and other stuff too... not sure what all exactly
21:53:51 <rbergeron> anyhoo
21:54:10 <rbergeron> I just wanted to feel out what people were thinking as we look towards the next round of fun.
21:54:43 <rbergeron> Anyone else have any thoughts atm?
21:54:47 <gholms|work> I think we should nail down the basics really well, first.
21:55:45 * rbergeron nods
21:55:57 <rbergeron> okee dokee.
21:56:00 <rbergeron> #topic Open Floor
21:56:10 <gholms|work> Hey, we have time for open floor this time!  :P
21:56:17 <rbergeron> I know!
21:57:34 <rbergeron> anyone? :)
21:58:04 <gholms|work> [A cat slowly rolls a watermelon up onto shore]
21:58:07 * rbergeron will close out the meeting in a minute or so if nobody pipes up
21:58:54 <jforbes> Thanks for participating everyone :)
21:58:56 <rbergeron> alllrighty then.
21:58:59 * rbergeron nods. Indeed!
21:59:06 <rbergeron> #endmeeting