18:59:58 #startmeeting Cloud SIG 18:59:58 Meeting started Fri Feb 10 18:59:58 2012 UTC. The chair is rbergeron. Information about MeetBot at http://wiki.debian.org/MeetBot. 18:59:58 Useful Commands: #action #agreed #halp #info #idea #link #topic. 19:00:03 #meetingname Cloud SIG 19:00:03 The meeting name has been set to 'cloud_sig' 19:00:17 #topic Welcome to the meeting, let's gather. 19:00:25 * rbergeron waits for people to pop in. 19:00:41 so when I come back on Monday am I going to have 500 emails about /usrmove -> about the future in my inbox? 19:00:46 Hai 19:00:54 kkeithley: yes 19:00:55 * rackerhacker waves 19:01:11 that's just wonderful 19:01:25 * ke4qqq shows up - but is in another meeting simulaneously. 19:01:39 Ditto. 19:01:48 Howdy 19:02:02 kkeithley: i could be proven wrong though 19:02:03 ;) 19:02:08 hey tdawson. :) 19:02:19 * jforbes is here 19:02:23 well then. I suppose it's time to pass out the hot dogs and get started 19:02:25 * gholms waves to everybody 19:02:39 hidey-ho to everyone! 19:02:46 * rbergeron wonders if dgilmore is around by chance 19:02:53 Are they Tommy's hot dogs, with extra chili? 19:03:02 They are delicious hot dogs, from Your Imagination. 19:04:00 #topic Feature-y statuses 19:04:03 OH LOOK. it's mull. 19:04:07 o/ 19:04:13 rbergeron: i am here 19:04:14 not to put all eyes on you.... 19:04:31 * rackerhacker mulls around 19:04:31 hi, rbergeron 19:04:31 dgilmore: will ping back shortly, just wanted to ask some ec2-ish stuff in a momento 19:04:36 rbergeron: watching the snow fall outside the window in the airport 19:04:39 mull: how is euca coming? :) 19:05:05 rbergeron, getting there... some other upgrade in rawhide threw me for a loop this week (jgroups 3 & jetty 8) 19:05:30 (There's a jetty 8?) 19:05:32 but I have got us building against groovy 1.8 now (instead of our bundled 1.7) which is good progress 19:05:38 * rbergeron saw your page got updated to a healthy 70% or .. something close to that 19:05:46 If you need me to do any reviewing let me know.\ 19:06:05 rbergeron, yeah, I updated to more accurately reflect the real state of things 19:06:22 gholms, please review wss4j ... should be simple 19:06:24 #info Eucalyptus is coming along; now building against groovy 1.8, some upgrades in rawhide caused a little blip; feature status is now appropriately updated 19:06:41 #info mull kindly requests gholms to review wss4j 19:06:46 gholms, how is the axis2c review? 19:06:52 * gholms gives rbergeron an #action 19:06:54 mull: Stalled. 19:07:00 spot is a busy man. 19:07:12 gholms: are you accepting? 19:07:14 ok, but it's in his court now? 19:07:24 rbergeron: shur-y-not 19:07:28 #info axis2c review is in spot's court 19:07:38 #action gholms to review wss4j as requested 19:07:40 :D 19:07:54 ke4qqq: how is cloudstack coming :) 19:08:27 * rbergeron wonders if sparks is around too 19:09:01 * rbergeron wonders if anyone else who is owning features is around but doesn't think so 19:09:33 * rkukura is here 19:09:53 * rbergeron notes casually that there were some people on the advisory-board mailing list who were lightly discussing packaging chef, if anyone's interested in that. 19:10:04 rkukura: can you talk about openstack statusy type things? 19:10:09 sure 19:10:21 #chair gholms 19:10:21 Current chairs: gholms rbergeron 19:10:44 I think essex-3 snapshots of most things, except horizon, are packaged now 19:11:06 #info essex-3 snapshots of most things, except horizon, are packaged now 19:11:15 * rackerhacker would like to start contributing a bit to the OS effort if help is needed 19:11:31 The linuxbridge plugin for quantum just got merged upstream today, so I'll update that package in the next few days. 19:12:13 The openvswitch kernel module is now being built, and the cdub is working on packaging the userspace tools. 19:12:15 #info rackerhacker is willing to pitch in and contribute a bit to the OS effort if help is needed 19:12:32 #info linuxbridge plugin for quantum got merged upstream yesterday, package will get updated in the next few days 19:12:45 i'm doing internal .deb packaging at rackspace for OS now but i'd like to get back to RPM's for my sanity ;) 19:13:02 #info openvswitch kernel module is now being built, cdub is working on packaging the userspace tools 19:13:04 I think that's about it 19:13:06 rackerhacker: LOL 19:13:17 * rbergeron sends rackerhacker her condolesnces 19:13:23 Heh 19:13:24 condolences 19:13:31 rackerhacker: lets chat offline 19:13:35 Seems like a good way to get the right dependency lists, etc. 19:13:44 rackerhacker: dprince feels your pain 19:14:33 lol 19:14:43 OKEEDOKEE THEN 19:14:51 * rbergeron still pokes at ke4qqq who is in a meeting 19:15:07 dprince knows where i'm comin' from ;) 19:15:45 rackerhacker: is it well described in a long poem by dante aligheri? 19:15:50 oh, wait, never mind ;) 19:15:56 * rbergeron moves onwards to ec2-ish things 19:15:58 rkukura: sure, we can sync up 19:16:01 #topic EC2 19:16:10 dgilmore: pingy now that i'm ready, lol 19:16:22 just curious if we were set to have Alpha/Beta going with ec2 19:16:30 or are we not doing that 19:16:38 * rbergeron assumes we are 19:17:01 seems like a reasonable thing to do 19:17:11 * mull is willing to test 19:17:31 rbergeron: ok 19:17:36 * gholms needs cloud-init testers, too 19:17:38 mull: very cool :) 19:17:47 rbergeron: im making some TC2 images now 19:17:47 gholms: can you... elaborate? 19:18:13 rbergeron: this is the firsttime that ive tried to make images since f16 ga 19:18:33 mull: speaking of testing - i know that there is going to be an openstack test day, are you planning on doing anything eimilar? 19:18:45 dgilmore: should it just work? 19:18:54 rbergeron: hopefully 19:18:54 or will it just break 19:18:58 rbergeron: we will see 19:18:58 okay 19:18:59 rbergeron, would love to if we can get the packages in soon enough 19:19:12 mull: *nod* gotcha 19:19:22 When you go to test cloud images, it would be useful to know what parts of cloud-init still need work, is all. 19:19:27 rbergeron: the image was created ok. ill upload to ec2 later and see if it actually boots 19:20:00 * dgilmore is also trying to make a qcow2 based image for people at home to use 19:20:07 #info on TC2 for alpha currently, dgilmore created an image, he will upload to ec2 later 19:20:21 dgilmore: can you let the list know when you did so people can poke around at it 19:20:34 dgilmore, I forgetare you using koji to create these images, or some other process? 19:21:58 rbergeron: yep will post amis when they are up and i confirm i can ssh in 19:22:04 mull: koji 19:22:15 mull: the tasks are running in koji now 19:22:48 * abento has an interest in testing cloud-init on ec2 images. 19:23:41 dgilmore, ok, I ask because I want to test processes that folks who don't have special koji access can use to create images (BG, ami-creator, whatever) 19:24:35 I will volunteer to test your images, too, though 19:26:20 mull: ok, well anyone can do it the way i am 19:26:33 mull: its just a kickstart and appliance-creator 19:27:04 dgilmore, oh, ok, I thought there were special privileges needed. 19:27:32 mull: there is in koji. but all its doing is calling appliance-creator 19:27:41 mull: anyone can do that themselves 19:27:41 gotcha 19:27:46 GAH INTERNET FAIL 19:28:10 #action dgilmore to post amis when they are up and he can ssh in 19:28:36 okay. anything else on ec2? 19:28:38 thanks, dgilmore :) 19:28:49 rbergeron: the ec2 image in qcow2 format is a 600mb image 19:29:23 dgilmore: can you exlain the usefulness of that for my edification 19:29:23 there has been quite a few requests to ship our ec2 images in a format people can easily use at home 19:29:46 #info ec2 images in qcow2 format are 600mb; there have been requests to ship ec2 images in a format people can use at home 19:29:48 rather than them installing they just spin up a vm with the image 19:29:49 ah 19:30:05 #info rather than installing they just spin up a vm with the image 19:30:07 i tarrred and xz compresses the f16 10gb raw image 19:30:07 dgilmore: I thought the interest was less "use at home" and more "upload to other cloud services" which makes qcow less useful 19:30:11 it was 114M 19:30:19 dgilmore: does that count as a fedora spin? not to open pandora's stupid box 19:30:34 jforbes: the requests ive personally seen have been for at home use 19:30:56 dgilmore: or perhaps the question is "how do we advertise those for use" 19:30:58 but i think we need to think about how we can ship something for use at home or in other cloud providers 19:31:17 cool, qcow2 is useful there, though so are tar xz compressed raw 19:31:29 rbergeron: i guess it kinda is a spin. rbergeron there was a proposal recently on the spins list for a minimal spin 19:31:37 dgilmore: would you be willing to send a mail to the cloud sig list to kind of start discussion on that thought (how we can ship something for use at home or in other cloud providers) 19:31:42 dgilmore: /me nods 19:32:02 jforbes: yeah the tar compressed raw is much smaller, but means people need to untar it after download 19:32:10 maybe we can make sparse raw files 19:32:36 rbergeron: sure i can send an email 19:32:40 dgilmore: thanks! 19:32:52 dgilmore: Either way. The storage migration feature in F17 will make it easy to migrate it to whatever storage backing you want 19:33:01 #action dgilmore to send an email to cloudsiglist to spur discussion on the "how we can ship something for use at home or in other cloud providers" 19:33:20 #info minimal spin is also a discussion on the spins mailing list 19:33:40 for the record the people i saw request it pointed me at http://uec-images.ubuntu.com/releases/10.10/release/ 19:34:02 #info requests came from people who were also pointing at http://uec-images.ubuntu.com/releases/10.10/release/ 19:34:39 How often would be update these images? 19:35:39 tdawson: i assume we'd just publish alpha/beta/final releases (someone can tell me if i'm wrong though) 19:35:47 How about once per (pre)release? 19:35:58 right 19:36:06 it would be alpha beta and final 19:36:21 (not nightlies, if that's what you're asking, tdawson) 19:36:28 And then we'd let the final sit until the next release comes around? 19:36:41 I was wondering if we'd have a monthly/quarterly update. 19:36:53 Definatly not nightly. 19:36:54 * mull guesses he was thinking "what if there's a remote root kernel exploit?" or something 19:37:04 tdawson: any particular reason you're thinking along these lines? 19:37:21 My last job (with Scientific LInux) I was asked to provide images. 19:37:33 And that was the discussion, how often would I provide them. 19:37:37 [Obligatory "GPL compliance" note] 19:37:45 just in case there's some duh use case i'm not aware of :) 19:38:43 In the end, nobody agreed on how often I would provide them, so I ended up not providing any images. :) 19:39:12 tdawson: LOL. which probably made you wonder just how important it really was to them if they couldn't come to an agreement :) 19:39:26 Yes 19:39:42 Proposal: once per (pre-)release. If someone needs them more frequently he or she is free to present a case for it. 19:39:51 Sounds good to me. 19:39:56 +1/-1? 19:40:02 +1 19:40:06 * gholms is +1 19:40:14 Anyone else? 19:40:24 * rbergeron looks at dgilmore and wonders if this works with how he might do other things 19:40:44 Umm, people can yum update the images anytime too 19:40:52 I'm neutral... I could be convinced that in the case of critical security flaws, we might occasionally spin an update 19:41:28 yum update doesn't fix the kernel on an AMI 19:41:47 It does when all we provide are EBS images. 19:41:56 ...with self-hosted kernels 19:42:28 Unless it is a remotely exploitable kernel issue that effects the select services in use on EC2, I don't think it matters much 19:42:35 gholms, you have to reboot though... are we going to tell everyone to update *then reboot* every image they launch? 19:43:10 there will only be one alpha, one beta and one GA images shipped 19:43:11 jforbes, understood that the case may be rare, but I wouldn't mind leaving the door open to say we _might_ _occasionally_ decide to spin an update 19:43:19 the same concerns could be said for livecds 19:43:25 Oh, absolutely. 19:43:43 That's what the "present a case" thing in my proposal is for. 19:43:51 Well, part of it. 19:43:56 mull, that's pretty standard practice anywhere you are doing kernel updates. LiveCD actually has more of a case for it since you are dealing with RO media 19:44:00 present a case == worst case scenario? :) 19:44:09 I agree with mull, but I also think that falls into the "present a case" ... I think remote exploit of kernels could be a case, depending on how bad it is. 19:45:02 ok, I think we're all on the same page... default is nothing after the GA image... except for the possible critical kernel thing... 19:45:27 So I think where we're at is ... alpha/beta/final (since that's the koji process and its built in already) and we can request the oh-crap fix in the event of a Bad Scenario and tackle it if we come to it, ie, "present a case" ? 19:45:54 Yeah, I think it's fine to say "we'll cross that bridge when we come to it" 19:46:07 And dennis will bring the more general topic of having the "minimal spin" type thing to cloud sig list. 19:46:11 yes? 19:46:18 * rbergeron looks around at everyone eating hotdogs 19:46:28 +1 19:46:45 WORKSFORME 19:47:13 * tdawson takes a bite of his Chicago Dog and give a +1 19:47:27 #agreed minimalspin/self-cloud can be alpha/beta/final, "present a case" if it's absolutely necessary and we'll figure it out if we have to at that point 19:47:29 * dgilmore should find the hotdog stand in ohare 19:47:46 dgilmore: isn't there one every like 12 feet? 19:47:55 ;) 19:48:00 rbergeron: i know of 2 hotdog places in the airport 19:48:02 * rbergeron looks around for other ec2-related stuff 19:48:09 * dgilmore has nothing 19:48:20 dgilmore: i'm glad to hear that you are indexing hotdoglocations in your brain :) 19:48:36 * rbergeron thinks "fedora dogs around the world" would be an excellent photo series 19:48:40 #topic Other Business 19:49:00 * rbergeron opens the floor for comments questions flames requests etc 19:49:12 * verdurin was impressed with the Fedora takeover at FOSDEM 19:49:24 verdurin: lol 19:49:57 there were quite a few folks there. 19:50:14 Do you have pix? :P 19:51:09 of fosdem? 19:51:16 In fact there were so many that talk clashes couldn't be avoided 19:51:18 In my imagination that i took on my imaginary trip 19:51:33 Hehe 19:51:43 * dgilmore wonders if anyone will be in Brno next weekend? 19:52:22 dgilmore: there will be lots of people, i don't think from here, though. 19:52:28 rbergeron: I'm going to Brno 19:52:33 not sure if any of the openstack crowd is going to brno, maybe rkukura is 19:53:04 i mean rkukura might be aware of who might be going 19:53:08 from that crowd 19:53:19 jsmith: ah, that's right 19:53:20 * dgilmore will be there and will be willing to talk to anyone about things 19:54:00 dgilmore: i won't be there, but you have fun. and jsmith too. 19:54:30 rbergeron: ill see if i can get you hotdog cloud pics :) 19:54:55 :D 19:55:56 [A tumbleweed drifts past] 19:56:11 [A robyn hugs gholms] 19:56:18 Alrighty, folks, thanks for coming. 19:56:21 Oh, yeah, congrats to you, rbergeron. 19:56:24 Keep those features humming along. 19:56:31 gholms: thanks :) 19:56:43 * rbergeron sets fuse and whatnot 19:57:03 see you all next week. 19:57:05 #endmeeting