21:03:09 #startmeeting Cloud SIG 21:03:09 Meeting started Thu Dec 16 21:03:09 2010 UTC. The chair is rbergeron. Information about MeetBot at http://wiki.debian.org/MeetBot. 21:03:09 Useful Commands: #action #agreed #halp #info #idea #link #topic. 21:03:10 rbergeron: sorry 21:03:17 #meetingname Cloud SIG 21:03:17 The meeting name has been set to 'cloud_sig' 21:03:24 #topic Roll call! 21:03:27 dgilmore: no worries :) 21:03:28 dgilmore: Three minutes is nothing. ;) 21:03:39 * jforbes is here and caffeinated :) 21:03:45 * mmorsi is here 21:03:49 Hai! 21:03:54 jforbes: woot for caffeine 21:04:03 hey mmorsi 21:04:05 hey 21:04:21 * gholms has also had way too much caffeine @_@ 21:04:25 clalance :) 21:04:26 * mmorsi drank a full cup of coffee and is still tired 0_o 21:04:28 * rbergeron needs moar caffeine 21:04:29 rbergeron: Sorry I'm late. 21:04:39 clalance: we are just getting started. 21:04:42 mmorsi: Thanks for the heads up :). 21:04:47 np 21:04:59 alrighty. 21:05:08 here here 21:05:09 Meeting schedule first? 21:05:11 #topic It's the most wonderful time of the year 21:05:21 cannot agree :) 21:05:27 mgoldmann: me either. 21:05:43 #chair gholms 21:05:43 Current chairs: gholms rbergeron 21:05:46 #chair jforbes 21:05:46 Current chairs: gholms jforbes rbergeron 21:06:03 #info It might actually *not* be the most wonderful time of the year. :-\ 21:06:03 So: I know a number of folks will be out on vacation during the next few weeks. 21:06:34 I wanted to see if we can continue to have cloud meetings, or if we will be too thin to do so. 21:06:41 I think we should be able to forge ahead. 21:06:43 http://www.youtube.com/watch?v=mvZ0KmPNAk8 21:06:46 its the cloudiest time of the year, how about that ;-) 21:06:49 gholms: did you ahve anything you wanted to add to that? 21:06:54 mmorsi: brilliant. 21:07:03 I will certainly be here next week, but not the week after 21:07:03 mmorsi: +1 21:07:11 rbergeron: I will miss the meeting on the 30th. Otherwise I can make it. 21:07:25 * rbergeron will be here next week, and may or may not be able to make it on the 30th, depending on what's going on. 21:07:35 How about this? Who can make it on Thursday the 23rd? 21:07:44 * rbergeron can make it on thursday the 23rd. 21:07:45 * clalance can 21:07:48 I will be here on the 23rd 21:07:56 * mmorsi can 21:08:02 * mgoldmann too 21:08:06 not the week after that tho 21:08:09 Sounds like a meeting is in order for the 23rd, then. How about the 30th? 21:08:15 * clalance cannot 21:08:19 * mmorsi can't 21:08:19 #agreed meeting will work on the 23rd 21:08:23 * jforbes will not be here on the 30th 21:08:28 * mgoldmann says maybe 21:08:34 * rbergeron says maybe as well 21:08:49 I'm moving, so I can't commit to either. :( 21:08:59 Okay. 21:09:23 Why do'nt we just cancel the meeting on the 30th, since it looks like we are pretty slow. 21:09:31 With the caveat that if anyone needs something, the mailing list is always open. 21:09:39 ya im moving onthe 30th too 21:09:40 ;) 21:09:41 And please raise your hand if you're having a fire, etc. :) 21:09:44 rbergeron: +1 21:09:46 works 4 me 21:10:31 #agreed cancelling the meeting on the 30th; please mail the list if you have anything needing addressing. 21:10:32 or call 911 21:10:35 Haha. 21:10:49 irc response times can be slow 21:11:06 thankfully 911 doesn't respond over irc. 21:11:11 :) 21:11:17 okay, moving forward! 21:11:22 #topic EC2 21:11:28 * gholms looks for brianlamere 21:11:30 jforbes: whatcha got 21:12:00 yeah, i don't see brianlamere about 21:12:09 :( 21:12:34 Nothing to report at the moment, I tried to get with dgilmore earlier today for the koji permissions and incantation to build images to test, but he asked that I contact him tomorrow. Infrastructure is busy today 21:12:46 yeah, they are. 21:12:58 They're going to be dealing with an overloaded netapp for a while, from the sound of things. 21:13:34 jforbes: coolio. 21:13:48 * rbergeron passes jforbes the "fastest ec2 wrapup ever in a meeting" award :) 21:13:52 Heh 21:13:53 netapp, schmetapp ;) 21:13:58 #topic Boxgrinder 21:13:59 I've got a small thing. 21:14:02 (for BG) 21:14:02 ohhhh 21:14:04 oh. 21:14:08 ;) 21:14:12 mgoldmann: how goes it? 21:14:18 very good! :) 21:14:23 gholms: ask away :) 21:14:35 thanks to gholms we have now working euca2ools 21:14:42 sweet. 21:14:43 We reportedly have images that work with EC2 now. 21:14:48 it needs karma :) 21:15:02 It's on its way to updates-testing, though when the next push will be is anybody's guess. 21:15:07 so, I'm going to push a full set of BG ASAP 21:15:14 gholms: 1.3.1-4.fc13, correct? 21:15:20 clalance: Yep 21:15:23 I need testers for packages that are already in updates-testing 21:15:32 So please get your packages from here for now... http://repos.fedorapeople.org/repos/gholms/cloud/ 21:15:33 brianlamere: welcome 21:15:38 greetings! 21:15:39 gholms: OK, cool. We are trying to update the deltacloud aggregator to the new BG. 21:15:46 So that should test it indirectly. 21:15:53 very good! 21:15:56 Then add karma at https://admin.fedoraproject.org/updates/euca2ools-1.3.1-4.fc14 and https://admin.fedoraproject.org/updates/euca2ools-1.3.1-4.fc13 , please. 21:15:56 (we'll add karma when/if we get it working) 21:16:04 That really should be when :). 21:16:08 holiday party happening around me, apologies for lateness 21:16:09 of course, perfect 21:16:31 rbergeron: so, moving forward, quite fast with BG 21:16:37 mgoldmann: Which version of the boxgrinder ec2 plugin uses euca2ools? 21:17:00 You need a nightly of the S3 plugin, IIRC. 21:17:09 OK, gotcha. Will pull that down. 21:17:10 it was not yet released, because I needed a working one 21:17:17 Ah, OK. 21:17:23 nightly will be build this night :) 21:17:47 where'd the kickstart file usage issue end up? 21:17:50 * gholms recommends depending on "euca2ools >= 1.3.1-4" 21:18:07 gholms: good point, adding 21:18:09 brianlamere: Here: https://issues.jboss.org/browse/BGBUILD-73 21:18:45 essentially mgoldmann has forked the kickstart file format by adding functional comments 21:18:58 >:( 21:19:19 exactly two: operating system name and version which kickstart file does not provide 21:19:58 many other plugns depend on those values (for example S3 - selecting proper AKI pair) 21:20:16 if you have better idea - I'm listening, really 21:20:22 Use pvgrub? 21:20:38 I'm thinking the ami-creator katzj posted seems like a good avenue for dev 21:20:38 we use pvgrub 21:20:49 Then you don't need to pick a AKI. 21:21:07 it's a vastly simpler base and therefore doesn't bring into it nearly as much feature requirements beyond the scope we actually require in fedora 21:21:27 well, partially correct, I have also support for fedora 11, but I think I get rid of that soon 21:22:29 f11 is EOL'd there's no need for supporting it from fedora's standpoint 21:22:43 #info ami-creator: http://lists.fedoraproject.org/pipermail/cloud/2010-December/000453.html 21:23:11 brianlamere: Any thoughts? 21:23:19 sure, if katzj's creator works better for Fedora rel-eng, I'm cool with that, BG is not meant for sysadmins really - it's a tool that helps other people: app developers, home users etc making live easier 21:23:43 skvidal: yes, I agree with the F11 bit 21:24:01 gholms: just was skimming over the issue - bookmarked it, I have a quiet weekend. everyone is leaving tomorrow, I'm not. I'll drop a line there 21:24:45 https://issues.jboss.org/browse/BGBUILD-117 21:25:12 mgoldmann: F12 is EOL as well. 21:25:30 * rbergeron looks around for anything else on BG 21:25:34 yup. but we don't support F11 on EC2 21:25:41 F12* 21:26:12 but yes, extending the issue 21:26:23 So do we know enough to decide whether or not we can use BG going forward? 21:26:46 One benefit we get is that it's more or less extensible to different guest formats so we can expand to other cloud providers. 21:26:58 we as in Cloud SIG/release engineering? 21:27:01 Yeah 21:27:10 gholms: I doubt releng is going to set it up 21:27:22 I think the answer we got from releng is BG will not be used at this time. That said it is a good tool for the community to build their own images 21:27:23 s/it/BG/ 21:28:05 jforbes: +1 21:28:23 So what is the plan as far as releng is concerned? 21:28:35 Can plain ol' appliance-tools build what we need? 21:29:03 In other words, it is important for Fedora rather releng uses it or not... Many people wanting to build their own images wont have koji access (not even sure how public that capability will be), and wont want to set up all of the koji -> publish scripting even if they do have koji access 21:29:25 gholms: it can provide the base image, not the EC2-ready one 21:29:28 ^ 21:29:48 gholms: plan for releng is using koji, which at the moment uses appliance-tools, but jeremy's project looks interesting there too 21:30:39 Then are they supposed to get a script that converts and uploads it somehow? 21:30:49 gholms: yes 21:31:20 well, ami-creator might take care of that part when it gets further along 21:31:46 out of curiosity - can someone dump out the requirement tree for boxgrinder? 21:32:08 repoquery --tree-requires 21:33:05 skvidal: You realize that's going to be huge, right? 21:33:13 Yeah, it's ridiculous. 21:33:15 gholms: which is why I was curious 21:33:23 (it goes all the way down to glibc dependencies) 21:33:37 clalance: then just get the first layer deep 21:33:44 we use libguestfs, this is the cause 21:34:08 skvidal: How does one limit depth? 21:36:03 * rbergeron wonders if we could take that to the mailing list? 21:36:09 Probably 21:36:17 sorry got pulled away for a omment 21:36:18 Anything else BG-related? 21:36:36 repoquery --tree-requires pkgname | grep '^ \\_' 21:36:41 that'll get top level only 21:37:00 if you have comments on kickstart file support, please add them here: BGBUILD-73 21:37:07 https://issues.jboss.org/browse/BGBUILD-73 21:37:23 #info if you have comments on kickstart file support, please add them to https://issues.jboss.org/browse/BGBUILD-73 21:37:51 Top-level isn't bad at all: http://fpaste.org/wPun/ 21:37:53 mgoldmann: bookmarked 21:38:08 skvidal: http://fpaste.org/eRGZ/ 21:38:13 Hm, although that is just for the s3-delivery-plugin. 21:38:17 I probably need to add more to that. 21:38:30 gholms: can you take that info to the mailing list? :) 21:38:40 rbergeron: I doubt it's necessary there. 21:38:48 clalance: try also ec2-plugin 21:38:49 I just meant so I could move forward :) 21:39:03 Sounds like there's no new BG stuff left. 21:39:08 yeah. 21:39:26 #topic CloudFS 21:39:30 jdarcy: any updates? 21:39:37 A couple. 21:39:44 There's actually some content in the git repo now. 21:40:02 #info CloudFS git repo has some content now 21:40:10 That's a plus. 21:40:19 That includes the "first drafts" of the recovery-logging, encryption, and multi-tenancy translators, with developer-level docs. 21:40:51 #link http://git.fedorahosted.org/git/?p=CloudFS.git 21:41:02 There's still plenty of work to be done on all three of those, plus real docs, plus setting up cloudfs.org with real content (other than the code). 21:41:11 it gets deep fast because it reqs the i686 and the x86_64 version of a couple things. 21:41:46 Still working with the Gluster folks on build processes so I can package this properly. 21:41:50 (sorry, that was delayed - typed it - pressed enter a few minutes later) 21:42:05 I think that's about it. 21:42:21 #info jdarcy is still working with Gluster folks. 21:42:24 excellent. thanks! 21:42:36 #topic Deltacloud & friends 21:42:43 jdarcy: lovin where it's goin, though - just fyi 21:42:51 brianlamere, thanks! 21:42:51 mmorsi, clalance: thanks for all the information you've been sending to the list. 21:42:56 np 21:43:00 You are welcome. 21:43:27 So I'm curious about what stuff, if any, will be coming to Fedora in the near term (meaning F15). :) 21:43:32 so yes, asked and the concensus is that we would like to add deltacloud as a feature for f15 21:43:46 I know there are some pieces - clalance, I know you had mentioned cloud engine as well - that aren't just deltacloud. 21:44:00 ah yes, things have undergone a bit of nemaing 21:44:01 mmorsi: do you know where all hte bits / information are to do that? 21:44:01 If you're going to add it as a feature for F15 would you please make it work with selinux? 21:44:03 renaming* 21:44:15 gholms: Yeah, that is on the list. 21:44:21 It should be a fairly simple matter, really. 21:44:23 will it works w/ it, but only if you have it permissive ;-) 21:44:33 (just haven't gotten to it yet) 21:44:35 but yes thats something to fix on the list 21:44:38 mmorsi: That isn't "working with it." :-\ 21:44:46 i was joking 21:44:52 Yeah, I know. :P 21:44:53 (hence the ;-) ) 21:44:55 rbergeron: With the fixed boxgrinder+euca2ools, one of the hurdles is out of the way. 21:45:08 Most of the dependencies are in fedora now (with the exception of a few). 21:45:31 clalance: is the deltacloud packaging page fairly up to date? 21:45:32 yes and we're making sure that any new ones that are added to the project are being added to fedora as we go 21:45:33 So it's a matter of packaging it up for F-15, and giving it a whirl to see what breaks. 21:45:38 rbergeron: Yeah, mostly. 21:45:41 Oh, hm. 21:45:45 http://deltacloud.org/page/Packaging_list 21:45:47 There is one other blocker for us, actually. 21:45:51 We require some custom condor code. 21:45:59 Which we have submitted upstream, but is definitely not in the Fedora package. 21:46:00 Ugh, condor? 21:46:02 Is getting deltacloud-client and deltacloud-core (Apache Deltacloud) into f15 a separate effort from the other deltacloud things? 21:46:08 gholms: Yeah, that's the scheduler we use. 21:46:13 gholms: ya we use it for our instance scheduling 21:46:18 instances are modeled as jobs in condor 21:46:19 bbrowning: Yeah, definitely separate. 21:46:19 I see. 21:46:24 gholms: you *love* condor, odn't you ;) 21:46:29 bbrowning: That's definitely something we can do right now. 21:46:40 rbergeron: Let's just say I have a hate-hate relationship with condor. 21:46:41 k - I was under the impression that getting deltacloud in f15 was just the apache bits, not the rest - good to know :) 21:46:53 bbrowning: Well, it is confusing with the name. 21:47:01 deltacloud API is the simple matter of doing a couple of packages. 21:47:06 The aggregator/engine is more complex. 21:47:27 So is the holdup there some patches to condor you have to get merged? 21:47:31 Yeah. 21:47:50 the packaging process could be started 21:47:50 Has to get merged upstream, and also then put into the fedora package. 21:47:54 the software wouldn't work 21:47:58 Unfortunately we require condor 7.5.5. 21:48:01 but would be ready to go 21:48:03 And I think fedora is still on 7.4. 21:48:08 (condor versions, that is) 21:48:33 Yep. Fedora only uses stable versions. 21:48:34 sdake: True. We do have packages already available, they just haven't gone through fedora review process. 21:49:25 clalance takes time to get those reviews going, best to queue them up so we know what we are faced with :) 21:49:42 sdake: Yes, true. How long do we have again until feature deadline? 21:49:57 jan 26 21:50:01 sdake: OK, thanks. 21:50:10 I'll try to enlist some help from the team for it. 21:50:20 At the very least, we can do more of the dependencies. 21:50:59 ha :) 21:50:59 ahem 21:50:59 Wow. 21:50:59 that explains where rbergeron went :) 21:51:00 Hopefully you can get your patches merged by the time 7.6 is released so you can meet all your deps. 21:51:06 Ouch. 21:51:10 Holy netsplit, Batman! 21:51:16 gholms: Yeah, we are shooting for that. 21:51:24 But we still have zodbot! 21:51:31 we won the kids 21:51:32 ;p 21:52:02 should prob put meeting on hold for few minutes while irc works itself out 21:52:06 * brianlamere hi-fives all the cool kids? 21:52:09 Si. 21:53:37 Schedule is here: http://fedoraproject.org/wiki/Schedule 21:53:40 So is condor likely to be your most problematic dependency? 21:54:11 what about boxgrinder? 21:54:20 What about it? 21:54:20 sgordon: Well, the new BG should fix that. 21:54:29 We are also going to move away from boxgrinder in the near future. 21:54:29 well it's the image builder... 21:54:34 and last i checked it isnt in 21:54:42 BG just doesn't quite suit your needs? 21:54:57 gholms: Not really, no. In particular, it can't do non-linux OSs. 21:55:01 Which is something we want to support. 21:55:02 Oof 21:55:30 * robynishere shakes the fist of doom at freenode 21:55:57 gholms: The other possibly problematic package is qpid/qmf. 21:56:04 We need at least a certain version of that. 21:56:09 the idea for BG was to write plugins for OS to support other OSes 21:56:09 No idea what version is currently in fedora. 21:56:24 * mgoldmann hides 21:56:30 mgoldmann: Yeah, understood. And we can write a boxgrinder plugin for oz that does that. 21:56:37 (I just haven't gotten around to it yet) 21:56:52 yup 21:57:00 wouldn't that be better than doing the work to shift to something else? 21:57:11 brianlamere: It's more or less the same thing. 21:57:25 clalance let me know which version of qpid/qmf you need, i'll ask aconway to address it in fedora 21:57:39 overall we need to decide if we are going to target dc for f15 or f16 21:57:47 there is a ton of work to do and f15 is fast approaching 21:57:51 sdake: OK. Right now it looks like we need: qpid-cpp-0.7.946106-3 21:57:53 with alot of dependencies in fedora out of date 21:57:55 Or better. 21:58:05 Right. 21:58:15 [Everyone falls silent and stares] 21:58:15 Ah, netreconnect :). 21:58:24 uhm 21:58:26 ok, what did you all miss? 21:58:33 HEY THERE PEOPLE 21:58:42 brianlamere: like, the last 15 minutes. 21:58:57 so summary is we are sorting out dependencies that deltacloud needs to be included in f15 21:58:59 rbergeron: We've been discussing cloud engine dependencies. 21:59:03 Thankfully we had zodbot here to help us out. 21:59:03 And where they are in fedora. 21:59:05 (I know, was the start of a joke that I...lost the punchline for. ha) 21:59:07 Ah yes. 21:59:23 brianlamere: woot 21:59:45 gholms, sdake, clalance: did we make notes on dependencies that need to get taken care of? 21:59:56 and by we i mean, did you make zodbot do that? :) 21:59:56 clalance: You might be better off holding off the feature until F16 for safety's sake. You can still add packages to F15 even without the feature. 22:00:04 gholms: Yeah, that's what I'm thinking. 22:00:10 gholms: We'll work towards getting more of our dependencies in. 22:00:17 And then we'll see where we are on the 26th. 22:00:23 (but as you can tell, we probably won't make it) 22:00:40 * rbergeron nods 22:00:41 Was there a page with DC deps on it? 22:00:49 gholms: Yeah, it's on deltacloud.org. 22:00:57 (it may be slightly out of date, but should be mostly right) 22:01:04 * clalance finds it 22:01:09 http://deltacloud.org/page/Packaging_list 22:01:13 There you go :). 22:01:37 Yeah, I need to add the qpid and the condor dependencies there. 22:02:13 #info Deltacloud needs a patched version of condor 7.5.5 to work properly; upstreaming is in progress 22:02:20 ah so the feature will be pushed back to f16 then 22:02:21 ? 22:02:24 that packaging list doesn't have the qpid/qmf dependency 22:02:28 Yeah. 22:02:29 prolly a good idea 22:02:31 I'll need to add it. 22:02:38 mmorsi: Yeah, we'll work towards what we can for F-15. 22:02:45 You can still add the stuff to F15 if you want. ;) 22:02:47 ah sure 22:02:49 Yeah. 22:03:10 somewhat unrelated but should look on adding a f15 feature for rails 3 and ruby 1.9 if we're gonna get both in 22:03:19 will definetly get the former in 22:03:21 not sure about the latter 22:03:23 mmorsi: I thought we were holding off on 1.9 until later? 22:03:23 but neways 22:03:25 i digress 22:03:33 OK, yeah, we don't have to do that here. 22:03:35 clalance: well if it gets in, it will be parallel installable 22:03:36 We'll talk elsewhere. 22:03:45 like python 2 / 3 22:03:46 sure 22:03:57 clalance: you'll add the qmf/qpid info to hte dc.o ackaging_list page? 22:04:01 gholms put action on me to ping Alan about an upstream update of qpid to atleast version qpid-cpp-0.7.94106-3 22:04:02 with the versioning needed 22:04:36 sdake: I updated the wiki page now. 22:04:43 clalance: thanks 22:04:52 #action sdake will ping qpid-cpp's maintainer about an update 22:04:55 are there any other deps missing we need? 22:05:04 sdake: There are, but they are mostly rubygems. 22:05:09 Those just need to go through the process. 22:05:23 sdake: 0.7.946106-4 has been built. Is it not pushed or something? 22:05:29 (at least, nothing sticks out to me as problematic) 22:06:05 qpid-cpp-0.7.946106-3.fc14 is in F14 stable. What is it you're missing? 22:06:05 when it doubt dont say anything 22:06:17 i have older version sorry gholms 22:06:21 #undo 22:06:21 Removing item from minutes: 22:06:21 gholms: Ah, yeah, that looks fine. 22:06:23 i run f13 :) 22:06:29 Ah 22:07:13 #info Sufficiently new version of qpid-cpp is incompatible with F13 due to soname bump 22:07:39 Ah, right. That's why we needed that (currently we are on F-13, we are moving to 14 soon). 22:09:05 [Somewhere in the distance, a dog barks.] 22:09:20 rbergeron: I think we are done with deltacloud for now. 22:09:36 yeh nothing to add here 22:09:46 Open floor / Open stack? 22:10:41 * gholms pokes rbergeron 22:10:42 I guess we lost rbergeron again. 22:10:49 #topic OpenStack 22:10:50 rbergeronn: ? 22:11:02 timing out? 22:11:03 I think they aren't the same people... 22:11:07 she fell through the open floor ;-) 22:11:08 Ah, oops :). 22:11:11 Yeah, maybe timeout. 22:11:13 Anything new on the openstack front? 22:12:11 Hm, I guess not. 22:12:23 What are we doing with OpenStack? 22:12:48 AFAIK there was a packaging effort. I'm not sure how well that has been going. 22:12:59 sigh. 22:13:09 * rbergeron lost her interwebz access 22:13:18 * gholms duct tapes rbergeron to the Internet 22:13:32 #topic Open floor 22:13:38 no kiddig 22:13:51 brianlamere: Any luck figuring out how to push stuff up to a S3 bucket? s3cmd? 22:14:10 * rbergeron will ping on the list about openstack - maybe silassewell can update us on that. 22:14:44 * rbergeron is typing from her droid, btw, so pardon any errors :) 22:15:09 I will absolutely have something posted on the list this weekend. my first open weekend in eons :P things have just been busy pre-holidays 22:15:34 funny that I can't just find 30 minutes to sit and focus on something... 22:15:59 I know what you mean. :-\ 22:16:29 #action brianlamere will mail the list his progress on the S3 mirroring front 22:17:15 Maybe I'm showing my complete ignorance, but...what's the status of eucalyptus in Fedora? 22:17:17 well I decided to take Xmas off - meaning, I am not running all over the country this year. just not gonna, darn it. quiet house, quiet office, if I can't find a 30min block somewhere then I'm simply broken 22:17:53 clalance: Eucalyptus bundles a ton of libraries, so it's blocked on separating those. 22:17:59 Ah, OK. 22:18:03 clalance: They're hiring me to, in part, fix just that. 22:18:08 euca2ools was the first/biggest thing, and that's in (right?) but....yeah, what gholms said 22:18:09 Cool :). 22:18:19 * rbergeron cheers for gholms 22:18:38 Oh right, since repetition is good: 22:19:02 Please test the euca2ools update to ensure that the images it builds work with EC2, then provide karma. 22:19:10