19:00:21 #startmeeting Cloud SIG 19:00:21 Meeting started Fri Aug 19 19:00:21 2011 UTC. The chair is rbergeron. Information about MeetBot at http://wiki.debian.org/MeetBot. 19:00:21 Useful Commands: #action #agreed #halp #info #idea #link #topic. 19:00:25 #meetingname Cloud SIG 19:00:25 The meeting name has been set to 'cloud_sig' 19:00:45 hey there, jgreguske. 19:00:49 #topic Who 19:00:51 rbergeron: afternoon 19:00:52 crap. 19:00:53 #undo 19:00:53 Removing item from minutes: 19:00:58 #topic Who's here? 19:01:02 * clalance here 19:01:05 * jgreguske here 19:01:13 * kkeithley is 19:01:15 hey clalance ;) 19:01:26 rbergeron: Hello :). 19:01:31 I'll just get started with a few updates while everyone rolls in. 19:01:34 #topic EC2 19:02:11 I saw jforbes this week, and he'll be able to get the F16 alpha images up on amazon using BG if necessary on Tuesday. 19:02:40 * quaid here on the slow side 19:02:44 #info jforbes will get F16 alpha images up on Alpha release day, using BG, assuming koji-made images aren't ready. 19:03:01 actually I've been uploading images a couple times this week 19:03:03 jgreguske: any updates on that front by chance? i haven't looked at the ticket in the past few days. 19:03:06 oh really :) 19:03:07 but I don't have a booting one 19:03:13 and I want to blame systemd 19:03:19 buuut I'm not really sure 19:03:36 I'm not blocked on a limitation of uploading or koji image building 19:03:36 ;) 19:03:51 well, no signs of those being the culprit rather 19:04:14 I'm not sure if the BG folks had anything they did for systemd-related stuff. 19:04:25 * gholms pokes his head through the door 19:04:36 there aren't any f16-alpha images built in BG as far as I know 19:04:45 so I don't have a basis for comparison 19:04:46 rbergeron: ive made images in koji 19:04:50 rbergeron: they dont boot 19:05:16 #info Progress on koji-built images: They're building, but not booting. dgilmore and jgreguske are investigating. 19:05:38 dgilmore, jgreguske: any details / bugs / etc. written in the rel-eng ticket by chance? errors or anything? 19:05:40 dgilmore: the ks files you use are the same that jforbes does? 19:05:44 maybe someone else might have an idea 19:05:53 dgilmore: I can post console output, sure 19:05:58 err rbergeron ^ 19:06:21 jgreguske: i used the f15 ks file from the cloud-kickstarts git repo 19:06:38 dgilmore: k, just want to make sure we're pulling from the same place 19:06:40 jgreguske: thanks. it may be helpful, or not, but on the chance it is :) 19:06:49 #link https://fedorahosted.org/rel-eng/ticket/4808 19:06:49 sure, understood 19:06:50 jgreguske: id like to see us append 3 to the grub boot line 19:06:59 just to make sure its booting to the right run level 19:07:13 #action jgreguske to post some of the details of failure in the releng 4808 ticket. 19:07:16 I will get BG images on F16 alpha this weekend, and we can see if we find anything there 19:07:23 hey justin :) 19:07:27 heya 19:07:30 Relatedly, I've been re-writing that upload script using python-boto 19:07:33 aren't you supposed to be vacationing? :) 19:07:41 driven with a config file it'll be much more distributable 19:07:43 partially 19:08:07 jgreguske: distributable only buys us so much when regular users cannot build images in koji 19:08:32 jforbes: they can build them locally 19:08:40 jforbes: using the exact same tools 19:08:46 #info jforbes is also working on re-writing the upload script using python-boto 19:08:58 rbergeron: thats jgreguske not jforbes 19:09:18 oh 19:09:21 #undo 19:09:21 Removing item from minutes: 19:09:23 you're right 19:09:37 #info jgreguske is also working on re-writing the upload script using python-boto 19:09:53 dgilmore: is image building in local koji installed documented anywhere? 19:10:14 not really. But you can use appliance-tools or Oz or whatever 19:10:34 jforbes: its just using appliance-tools 19:10:52 jforbes: anyone can run that locally in mock and make an image 19:11:10 jgreguske and dgilmore: thanks for your help on this. i am assuming you guys are not expecting it to be solved by tuesday, so I think the plan at this point is still to just have justin make them with BG 19:11:12 they don't even need mock if it is for their own play 19:11:17 or just run it locally and use the upload scripts 19:11:36 Cool, once the scripts are redone, we should add it to the "make your own images" page 19:11:42 sure 19:11:52 rbergeron: considering that jforbes hasnt even attempted a f16 image i dont expect that there will be anything on tuesday 19:12:16 jgreguske: right 19:12:17 dgilmore: shouldnt be a problem 19:12:20 Just curious, what does the boto-using script do that euca2ools doesn't? 19:12:35 Trying to avoid duplicated effort here 19:12:36 dgilmore: it only takes an hour to go from launch to published image, so I can go through a lot more debug cycles 19:12:46 jforbes: are you doing something to set the default run level for systemd? 19:13:11 gholms_: the difference is that you can put some standardization around it 19:13:24 dgilmore: not specifically, though I thought BG picked that up for F15 19:13:44 Oh? 19:13:44 gholms_: make sure you always register them with the same block mapping, same security group, AKI, etc 19:13:57 gholms_: usign the same credentials, accounting 19:14:03 that's all really 19:14:27 gholms_: also, euca2ools doesn't upload to image to EC2 from a file afaik 19:14:35 no EC2 API for that 19:15:04 msavy: any thoughts on systemd with bg? was that all enabled for F15? 19:15:07 Interesting; I shall have to look at said scripts. 19:15:24 yep, standard systemd setup for our F15 images. 19:15:47 for F16 they've now moved onto GRUB2, so we'll have to make some changes to build those images 19:15:59 jforbes: im kinda suspecting that some of our issue is taht the default boot target is wrong from appliance-creator 19:16:05 but thats a bit fo a guess 19:16:54 jforbes: do you think the grub2 stuff will prevent building stuff with BG by chance if it's not sorted through yet? 19:17:42 rbergeron: Likely could. We have a problem in that we dont actually use grub in EC2, we use the grub.conf, and it only understands the old grub.conf 19:17:55 rbergeron: though since we don't actually care if grub2 works, that can be worked aroudn 19:17:59 okay. well, food for thought, then, i suppose. 19:18:54 I just pasted console output in that ticket 19:19:06 ugh and the formatting is terrible! 19:19:07 jforbes: huh 19:19:07 * rbergeron wants to move on a bit, since we have plenty to cover today - can we keep up any discussion re: getting things working either in the ticket or mailing list? :) 19:19:40 someone mentioned systemd targets, in BG we twiddle that using libguestfs https://github.com/boxgrinder/boxgrinder-build/blob/master/lib/boxgrinder-build/plugins/os/rpm-based/rpm-based-os-plugin.rb#L196 19:20:14 and we use pvgrub for EC2 builds, which is probably what you are referring to jforbes? 19:20:41 rbergeron: silence == consent ? :) 19:20:42 msavy: right 19:21:09 msavy: pvgrub is the AKI that amazon provides, it uses our grub.conf, it doesnt care about our grub 19:21:37 yep 19:22:28 okay, /me is moving on 19:22:37 #topic Aeolus 19:22:48 clalance, mmorsi: HI! 19:22:57 rbergeron: Hello :). 19:23:02 howdy 19:23:08 how's things? 19:23:15 rbergeron: Hm, let's see. I guess not a whole lot. 19:23:22 Development is ongoing on our next sprints. 19:23:35 We've moved the whole development team to F-15 and dropped F-14 for now. 19:23:40 * rbergeron notes that further development Aeolus for F14 appears to be going byebye, in case anyone is concerned. :) 19:23:44 JINX, of sorts 19:23:47 Haha. 19:23:48 Yeah. 19:23:54 I don't think anyone will scream. 19:24:15 And we are working on our next sprint. 19:24:22 We'll probably have an interim release next week sometime. 19:24:27 (at least, I hope we will) 19:24:51 Cool. :) 19:24:59 And that may be what I finally package for F-16. 19:25:00 * rbergeron would like to advertise something real quicklike, as well 19:25:13 rbergeron: Yeah, I know nobody responded to your request for video. 19:25:14 * clalance is shy 19:25:16 nothing from myself other than to just look for a screencast / getting starting w/ aeolus on fedora guide next week 19:25:24 clalance: rbergeron i did 0_o 19:25:28 #link http://www.youtube.com/aeolusproject 19:25:33 mmorsi: yes, you did :) 19:25:38 mmorsi: Ah, OK, I didn't see it go by. 19:25:40 mmorsi: Thanks :). 19:25:46 And I got a note about a screencast someone did as well, linked above. 19:25:49 clalance: np, was a private response 19:26:01 Oh, I do have a question for the wider group 19:26:05 #info Check out the aeolus screencast if you're curious about WHAT IT IS 19:26:06 (I already asked rbergeron this) 19:26:08 sure, go ahead. 19:26:11 Yeah, i probably forgot. 19:26:17 ah ya, my screencast is going to have the complete tooling 19:26:20 Oh, the bodhi q? 19:26:21 Why is it that all of my F-16 updates are stuck in testing? 19:26:23 aeolus-image, deltacloud, conductor, etc' 19:26:25 rbergeron: Yeah. 19:26:35 That page you pointed me to wasn't that enlightening. 19:26:35 clalance: have they been marked as stable? 19:26:38 mmorsi: Yes. 19:26:48 hrm when did you mark them as such 19:26:54 mmorsi: Some of them weeks ago. 19:26:57 * ke4qqq shows up late 19:27:12 hrm weird 19:27:23 might be worthwile to file an issue w/ release engineering 19:27:41 dgilmore: Any thoughts? 19:27:46 the f16 updates were the ones marked as stable right? 19:27:49 not f15 or w/e 19:27:53 dgilmore: Are we in some sort of freeze period? 19:27:55 mmorsi: Yeah, exactly. 19:28:28 There simply have may yet to be a stable push post-alpha. 19:28:34 clalance: yeah, we are in freeze :) 19:29:01 rbergeron: But I thought we could build new packages up until beta? 19:29:05 (that's where my confusion is) 19:29:59 clalance: so when we're in alpha freeze, my understanding is that's up until alpha goes out, or perhaps at least until we have the right RC. 19:30:07 rbergeron: Ah, OK. 19:30:12 So that might explain it. 19:30:18 so we aren't having random things break, only critical fixes from blockers 19:30:22 All right, I'll wait until after Alpha is released. 19:30:32 But um, while I should know, perhaps someone smarter could pipe in. 19:30:50 rbergeron: Heh, OK. Well, we don't have to hold up for this, I'll open a ticket and they can tell me where I am wrong :). 19:31:03 okay. 19:31:32 gold != alpha? 19:31:41 I think that is all from Aeolus. 19:31:54 There's a clown in every (chat) room. 19:31:55 ;) 19:32:04 clalance: sorry reading abck now im not paying attention to the meeting 19:32:18 clalance: yes updates are frozen 19:32:30 only alpha blockers have gone to stable 19:32:37 dgilmore: Ah, OK. That's fine, but presumably they will be unfrozen before beta again? 19:32:50 now alpha is done updates will flow to stable until we freeze for beta 19:33:02 dgilmore: OK, perfect. That's what I wanted to know. Thanks. 19:33:21 kkeithley: so the RC was delared gold, and that's basically what alpha IS, but we don't "officially release" it as Alpha till tuesday. 19:33:36 mirrors and docs and trans and marketing stuff all needs to happen in the interim, etc. 19:35:44 okay 19:35:45 sorry. 19:35:47 * rbergeron moves on 19:36:16 #info Aeolus in F14 is coming to an end, as an FYI. 19:36:22 #topic HekaFS 19:36:28 :( 19:37:07 gholms_: Why the long face? 19:37:22 Loss of f14 support 19:37:36 I was on vacation for a good chunk of the past week, so not much to report. Mostly focus on testing. 19:37:39 (for Aeolus) 19:38:18 Good news is I've been running tests on two clients connected to two servers, specifically to check for cross-tenant interference and memory leaks and such. Over 1M iterations on each client and things are still running fine. 19:38:38 Nice! 19:38:52 #info jdarcy is running tests on two clients connected to two servers, looking for "bad things happenin", has 1M iterations on each client and they're still running fine! :) 19:39:08 * kkeithley some testing, and mostly whining/whinging and nagging to get commit privs for glusterfs so I could update glusterfs to the latest stable upstream (3.2.2). Success there, and both glusterfs and hekafs pkgs are in the update queue 19:39:11 So no news is good news, I guess. HekaFS is basically looking good for F16, though we'll stay quite busy for a while. 19:39:17 * rbergeron nods 19:40:08 kkeithley: Are you planning to update beyond 3.2.2 for F16? 19:40:19 #info kkeithley has glusterfs in fedora updated to the latest stable upstream, glusterfs/hekafs pkgs are in the update queue. 19:40:35 kkeithley: I have some patches to make the parts we cloned build against a 3.2.3-ish tree. 19:40:36 I will if gluster puts a newer tarball on .... 19:40:51 http://download.gluster.com/pub/gluster/glusterfs/3.2/ 19:41:18 unless we decide on some other source, but that's what's been used thus far 19:41:44 Did you manage to hear from the glusterfs folks? 19:41:45 That's the right source, I see no reason to change. 19:41:58 gholms_: hear from them .. you mean the maintainers in fedora? 19:42:14 daMaestro was in #gluster earlier today. 19:42:27 Didn't get a chance to talk, though. 19:42:39 gholms: yeah, which glusterfs people, fedora or Gluster? 19:43:01 I filed a bz and thias gave me the commit privs I needed, if that's what you mean 19:43:13 Oh that was it. 19:43:16 Cool 19:43:28 Yeah, it was the maintainers. 19:43:45 yup. 19:43:50 BTW, I might use some of my "free time" to get GlusterFS/HekaFS working on my new Android tablet. :) 19:44:04 #info kkeithley has obtained maintainership status for glusterfs in fedora also 19:44:07 O.o 19:44:14 Alllllllllllrighty then. 19:44:17 Thanks, guys. :) 19:44:23 #topic OpenStack update 19:44:30 So for those who aren't up to date in their mail - 19:44:43 #info markmc sent a nice mail to the list about OpenStack packaging status 19:44:56 #link http://lists.fedoraproject.org/pipermail/cloud/2011-August/000733.html 19:45:12 I propose a motion to commend markmc for general awesomeness. 19:45:21 #info He's put together a nice list of "to-do's" and there are a number of packages that just need review that he's got put together in a freaking awesome wiki page. 19:45:32 #link https://fedoraproject.org/wiki/OpenStack 19:45:34 jdarcy: srsly. 19:45:44 #agreed markmc is commended for general awesomeness 19:45:46 * rbergeron grins 19:45:51 jdarcy: +1 19:45:53 :) 19:45:58 Heh, definitely. 19:46:19 Anyway: if anyone's interested, as always... 19:46:57 rbergeron: I took a swag at reviewing one of the packages. 19:47:05 That was about all I could bring myself to do today. 19:47:13 #info Feel free to pitch in and help with reviews! 19:47:32 #link https://bugzilla.redhat.com/707199 19:47:43 #link https://bugzilla.redhat.com/731966 19:47:54 #link https://bugzilla.redhat.com/731980 19:48:14 #info Glance, Nova, and novaclient packages ready 19:48:21 clalance: awesome. thank you :) 19:48:50 #topic Other business 19:49:18 No BoxGrinder ;-)? 19:49:28 #info pacemaker-cloud and sheepdog Features for F16 are in good shape/done-ish. 19:49:31 did anything happen with BG?? 19:49:34 #link https://fedoraproject.org/wiki/Features/pacemaker-cloud 19:49:35 :) 19:49:48 msavy: crap, I suck. hang on :) 19:50:03 #link http://fedoraproject.org/wiki/Features/Sheepdog 19:50:16 * rbergeron also notes it's time to really actually start srsly thinking about test days for all these features. 19:50:33 There was just a new BG release, right? 19:50:35 #action rbergeron to send harassing yet loving note to mailing list about Cloud test days for some of our features. 19:50:40 yep! 19:50:46 * ke4qqq notes that he hasn't received any feedback on the swift 1.4.1 packages and will consider silence golden and is going to push to F16 and rawhide in the not too distant future. 19:51:10 #chair gholms_ 19:51:10 Current chairs: gholms_ rbergeron 19:51:33 #info ke4qqq hasn't received feedback on swift 1.4.1 packages and will prob push to f16/rawhide in not too distant future 19:51:53 * quaid has a topic 19:52:04 #info New BoxGrinder release yesterday 19:52:09 #link http://boxgrinder.org/blog/2011/08/18/boxgrinder-0-9-4-released/ 19:52:25 Darn it, stole my thunder ;-) 19:52:32 * rbergeron hugs msavy 19:52:45 #chair msavy 19:52:45 Current chairs: gholms_ msavy rbergeron 19:52:54 #chair quaid 19:52:54 Current chairs: gholms_ msavy quaid rbergeron 19:53:00 quaid: What is it? 19:53:04 * rbergeron is going to hand off to quaid for a second, he has something he wants to bring up 19:53:05 #link https://fedoraproject.org/wiki/CUDCon_proposal 19:53:18 I want to do an alongside-of-FUDCon cloud-focused event 19:53:45 partially because - to be honest - I can't easily fit OpenShift talks in to FUDCon without at least an off-topic-track :) 19:54:13 #info CUDCon proposal page explains full reasoning why a new stand-alone conference brand built on the FUDCon brand/model 19:54:25 * rbergeron looks at ke4qqq for thoughts, since he's cloudy and also has been to fudcons 19:54:31 I was thinking of starting with a side-track in Milan 19:54:42 milan - wow thats soon 19:54:43 quaid: I think it would be worth also proposing to the Milan folks. 19:54:45 and I have at least Tobias Kunze from OpenShift interested in making it happen 19:55:00 ke4qqq: and then we learn, and that helps with FUDCon NA :) 19:55:07 rbergeron: +1 19:55:11 Because I do'n tknow what they have available in terms of space, or if they're wanting to share that time - 19:55:19 #action quaid to propose CUDCon to FUDCon Milan proposers 19:55:36 quaid: open to participation from others or is it openshift focused? 19:55:37 A lot of times we see things where FUDCons are colo'd with other events, and people have a hard time choosing what they want to do 19:55:38 "Yo, dawg" 19:55:41 ke4qqq: others, for sure 19:55:50 ke4qqq: OpenShift is just the example 19:55:55 * ke4qqq if I come I'll bring some sponsorship with me...../me guesses same is true for gholms peeps 19:56:04 quaid: but open source focused? or do they have to have a package in fedora, or ... ?? 19:56:05 ke4qqq: others may say,"Oh, we're not in Fedora, FUDCon isn't there for us" 19:56:10 It's an interesting idea for sure. 19:56:28 #info "Bring in open community cloud efforts of all backgrounds that are built on an open source stack that include Fedora or its remixes/downstreams such as RHEL, CentOS, etc. " 19:56:32 quaid: does tobias have cashola to back this up? 19:56:34 I like the idea. 19:56:44 can we toss this idea on the mailing list? 19:56:56 rbergeron: maybe not for Milan, but that is part of my plan for NA 19:56:57 yeah, I really do wnat to cross-post it with the milan folks as well. 19:57:00 quaid: yeah 19:57:03 Milan is also a cheap proof-of-concept 19:57:16 to sell the marketing idea to *not only Red Hat* 19:57:16 When is that event again? 19:57:18 yeah i don't know if I can do milan, and our .eu presence is pretty weak at the moment. 19:57:36 ke4qqq: well, let me prove the idea for you, if it works, you can propose working in presence for NA 19:57:37 quaid: BTW, I have had fascinating convos with Kristin this week, we might pull her in as well. 19:57:51 My wife still chuckles at "FUDcon" so I'm sure "CUDcon" would be a big hit. 19:57:53 gholms_: FUDCon NA is Feb 2012 (right?), Milan is 30 Sep to 02 Oct 2011 19:58:21 rbergeron: Nordstrom? 19:58:29 ok, I'll send proposal to the list 19:58:42 I just wanted to float the idea while we were here for Q&A 19:59:03 jdarcy: CUDCon, "Ruminate with us about clouds!" 19:59:04 * rbergeron nods 19:59:08 quaid: yes 19:59:22 jdarcy: we need a cow logo i think - we can make it part of the meat sig 19:59:30 ke4qqq: bingo 19:59:34 ... 19:59:38 Cloud. It's what's for dinner. 19:59:41 ke4qqq: does that idea feel inclusive to you of e.g. Cloud.com efforts? 20:00:06 quaid: had a big talk with her about rht presence at events, and we're going to just start a wiki page, and i'm going to coordinate with her about getting notices out about calls for papers, and basically try and prove to People with Money that community events are worth putting more cash into and presence into. 20:00:09 as in, would you think, "Oh, yeah, that's a place where I can interact with other upstreams and also get a leg-up oni Fedora/RHEL/CentOS/SL connections" 20:00:09 quaid: yes - milan might not just too short a notice, but blacksburg does. 20:00:25 ke4qqq: yeah, that's OK 20:00:37 quaid: but I'd want the virt guys there as well if I had the magic wand.... :) 20:00:37 quaid: and I think she'd love to help out with this type of idea as well 20:00:41 I mean, that's more than I could ask for! 20:00:48 ke4qqq: that may be arranged :) 20:01:05 ke4qqq: I think there is a fair chance that we could see KVM and management of virt and stuff there 20:01:10 ke4qqq: they'll honestly be easier to tap in Milan 20:01:19 and if they still let me have the wand, maybe the BG folks, puppet, pulp, etc. 20:01:24 ke4qqq: would <3 to see Open Virtualization Alliance members coming 20:01:48 * gholms_ will see if he can send someone 20:01:56 as long as it stays technical i'd agree. 20:01:56 alright. 20:02:07 * ke4qqq doesn't want it to turn into a glossy hand out fest. 20:02:15 quaid: can you take the action to post thoughts on the mailing list for cloud, and maybe also put in a line to the fudcon mailing list? 20:02:41 ke4qqq: I think generally having a more "community" cloud/virt fest that is more authentic and is dev-focused would be FREAKING AWESOME. 20:02:46 * jdarcy wonders if the Gluster folks would be interested. 20:02:48 #action quaid to post to cloud@ about CUDCon 20:02:55 rbergeron: already did an action about contacting Milan folks 20:02:58 jdarcy: I'll go downstairs and ask them 20:03:02 jdarcy: /me thinks John Mark certainly would be 20:03:04 and ke4qqq can harass them further 20:03:07 and you can as well :) 20:03:15 quaid: you rock 20:03:16 * jdarcy nods. 20:03:22 * quaid is done for now 20:03:28 quaid: thanks! 20:03:36 cheers! 20:03:54 ke4qqq: remind me to harass you about build-a-open-source-cloud-day and if there are any open spots for any of our beloved projects here :) 20:04:21 ok 20:04:31 darn, that's right, SCALE10x 20:04:35 ok, I'll talk about that next time :) 20:04:40 quaid: what about it? 20:04:57 I want to do OpenShift, Cloud, Fedora,Red Hat stuff there :) 20:05:00 oh shit. 20:05:04 is that the same weekend as fudcon? 20:05:09 ? 20:05:23 oh, no, it's not. 20:05:24 whew 20:05:28 they moved it back :) 20:05:32 * quaid runs for real now 20:05:33 yeah 20:05:35 Okay. 20:05:38 I'll bring it up next week, don't let me forget 20:05:38 Anything else, anyone? 20:05:41 or in between 20:05:44 * quaid & 20:05:44 * rbergeron grins 20:05:54 * rbergeron salutes everyone for a lively meeting this week 20:05:55 [The Easter bunny flies through the window] 20:06:00 HOLY CRAP 20:06:02 * rbergeron looks 20:06:04 i missed it! 20:06:52 okay. 20:06:57 Thanks for coming, folks. 20:07:03 #endmeeting