18:59:28 #startmeeting Cloud SIG 18:59:28 Meeting started Fri Nov 18 18:59:28 2011 UTC. The chair is rbergeron. Information about MeetBot at http://wiki.debian.org/MeetBot. 18:59:28 Useful Commands: #action #agreed #halp #info #idea #link #topic. 18:59:39 you made it 18:59:56 #meetingname Cloud SIG 18:59:56 The meeting name has been set to 'cloud_sig' 19:00:17 Ooh, you're early. 19:00:19 #topic Who be here? 19:00:36 * rbergeron runs inside with her food 19:00:38 * kkeithley is 19:00:43 * mdomsch 19:00:57 chirp, chirp 19:01:01 is this the meteorology club meeting? 19:01:48 word 19:01:52 I think so. 19:01:54 jforbes: Today's forecast: this meeting will be cloudy, followed by clouds. 19:02:10 #topic Tell me you have something exciting to talk about. 19:02:12 with a chance of meatballs? 19:02:19 #chair gholms jforbes kkeithley mdomsch 19:02:19 Current chairs: gholms jforbes kkeithley mdomsch rbergeron 19:02:22 rbergeron: we have an EC2 item to discuss 19:02:24 * mull is here 19:02:24 OH! 19:02:26 Awesome. 19:02:28 * jsmith wants more cloud -- this burning ball of gas in the sky makes his head hurt 19:02:28 Hey mull ;) 19:02:35 #topic EC2 19:02:40 take it away, sir 19:02:47 while i get plates and utensils 19:03:10 We have our first encounter with EOL on EC2 images, and I think we need to come to an agreement on the best way to handle this 19:03:44 hasn't F8 been eol'd for years? :0 19:04:02 rbergeron: We didn't publish those. 19:04:03 F14 images are currently supported, and soon will not be. Our options are 1) Delete the reference from the "supported images" page 2) Delete the reference and make the images non public 3) delete the reference and the images 19:04:09 rbergeron: we have no control over those 19:04:49 * jsmith leans toward option 2 19:04:50 I would probably recommend 2, simply because if enough uproar is raised, we can make them public again until a better solution is found 19:04:53 I think it'd be better to make the image non-public, but still have it ... you know, in case of emergency or reference 19:05:01 that too 19:05:12 Just like our regular images -- we're not going to actively force someone not to use them... 19:05:26 Making the images non-public will make it so people cannot use them. 19:05:27 If you know where to find it and you really want to use that version... 19:05:36 Hmmmmn... 19:06:01 jsmith: right, making them non public makes them unusable, but if people complain enough we have the option of making them public again. 19:06:04 (Just FYI) 19:06:27 gholms: ohhh 19:06:53 jsmith: but if they are public, they are visible on search results, which doesn't let the non Fedora familiar know that they are completely unsupported either 19:07:13 jforbes: OK, I guess I'm still leaning towards option 2 then 19:07:38 Slightly less thrilled with it, but I still think it's the best option 19:07:48 (and we can always change our minds later if we find the decision untenable) 19:08:12 jsmith: I agree, it would be nice if we could perhaps make them hidden but usable for people who knew where to look, but I don't know of any option there 19:08:35 Is anyone against option 2 or think that another option might be better? 19:09:22 Cry? 19:09:25 Ah, well. 19:09:30 okay, I will handle deactivating them on F14 EOL 19:09:34 At least it may be hte moment where we hear people complain ;) 19:10:01 Complaining just means we have an opportunity to teach and improve :-) 19:10:04 #action jforbes to deactivate F14 images on EOL 19:10:06 In the future, we might want to come up with an option that lets the users know on boot when an image they are using is no longer supported. I need to think on that a bit 19:10:27 * rbergeron nods 19:10:46 maybe cloud-init could check release against a file somewhere with supported releases listed and warn? 19:11:13 which would give us some added metrics as to how often official Fedora images are booted in EC2 19:11:20 That is definitely possible. 19:11:39 IIRC, Ubuntu images query a central server to print the latest version of an image. 19:11:56 jforbes: ooh, though that also makes me "oooh privacy" and stuff too 19:12:09 * jsmith sees alarm bells go off in his head 19:12:18 IP addresses and such are not relevant in EC2 though, all we could really gather is access count on the file 19:12:19 though it's not that invasive 19:12:22 right 19:12:26 jsmith: Eeeyup 19:12:39 I have no problem about checking version, but the "added metrics" has me a little more concerned 19:12:52 it's hot many times the file was accessed 19:12:52 (again, depending on how it's done, it could be perfectly fine) 19:13:02 That I'm fine with 19:13:14 (just like we handle yum repodata and direct download statistics today) 19:13:18 right, no additional probing or anything, that would be wrong 19:13:23 PROBING 19:13:24 WORKSFORME 19:13:38 +1 19:14:13 gholms: so perhaps the check could look online for a file, the file will list the 2 most recent supported versions and perhaps give a warning and a link to our supported image page when found to be out of support 19:14:34 Like I said, this wont help us with F14-16 EOL, but it will with F17+ 19:14:50 Aye... 19:14:58 That sounds like a good starting point. If it needs to get fancier than that we can figure that part out later. 19:15:28 * jforbes certainly doesnt want to get too fancy and bring up privacy issues. 19:15:28 gholms + jforbes: so which one of you is going to start on makign this magic happen? 19:15:28 Let's get version 0.1 out there, see how it goes, and iterate from there 19:16:20 -ENOTIME 19:16:30 gholms: yeah, i know 19:16:34 I am not particularly familiar with the cloud-init code, but it should be fairly simple to do. I cna take a look 19:16:43 jforbes: take a peek and maybe let us know? :) 19:16:49 will do 19:16:49 maybe someone else might be itnerested in diving into it. 19:17:04 jforbes: Just read a couple plugins in cloudinit/CloudConfig; it's relatively straightforward. 19:17:05 * jforbes is happy to defer if anyone else steps forward 19:17:12 #action jforbes to look into adding EOL information to future releases, report back to list 19:17:29 jforbes: maybe a cursury look at "what to do" (or gholms could maybe pipe up a bit) would be helpful 19:17:36 without going ... crazy 19:17:43 might give someone else the thought that they could do it 19:17:50 rbergeron: sure 19:17:52 whatever works ;) 19:18:07 * rbergeron looks around for any other ec2 stuff 19:18:18 rbergeron: though from what I am thinking, it will probably take less time to write the code than to write the email explaining it 19:18:24 jforbes: :) 19:18:25 I just want to again say thanks for having EC2 images available on GA day 19:18:30 jforbes: If you have questions feel free to ask. 19:18:31 do we have images up in europe? 19:18:32 That made me very happy :-) 19:18:35 rbergeron: We do! 19:18:40 gholms: thanks 19:18:48 okay, i oculdn't remember 19:18:53 i saw someone asking about it the other day 19:18:55 rbergeron: Or, at least I thought we did 19:18:56 rbergeron: we have images in *every* availability zone, including the one they launched right after F16 GA 19:18:59 elsewhere 19:19:11 Yes, we have images in eu_west region 19:19:15 jforbes: yeah, i saw dennis and max going back and forth 19:19:25 We should be good :-) 19:20:05 That for uploading and testing it, you two. :) 19:20:22 * rbergeron looks at gholms' sentence and moves onwards 19:20:29 #topic FUTUREFEATURES 19:20:35 So, let's see here: 19:20:37 A few things 19:21:03 #info I have a few people inquiring about packaging oVirt and "how-to" (and they're not even packagers yet)... so I may be buttering people up here soonish. 19:21:31 #info Wanted to check in with mgoldmann on how AS7 is going, but I'll see if he's around Monday 19:21:38 * jsmith points out that he'd like to have the "Gloud Guide" as a feature for F17 19:21:50 rbergeron: Last I checked, there were six or seven packages still waiting to be reviewed 19:21:51 jsmith: you are welcome to submit a feature page 19:21:54 jsmith: Gloud Guide? 19:21:57 rbergeron: I may just do that :-) 19:22:09 jdarcy: http://docs.fedoraproject.org/en-US/Fedora_Draft_Documentation/0.1/html/Cloud_Guide/index.html 19:22:21 jdarcy: That's the version for GlusterFS. :P 19:22:44 rbergeron, re: AS7 ... I think I probably owe him some package reviews 19:22:51 * jdarcy makes a note to contribute some bull^H^H^H^Hmaterial. 19:22:54 not much movement on his wiki age lately 19:23:14 mull: okay. how are we coming along with Euca-stuff? 19:23:22 Now is a great time to beg for reviews and help. ;) 19:23:34 rbergeron, I got two packages approved this week 19:23:42 mull: yay 19:23:54 and one more reviewed but waiting for final approval 19:24:06 (translation: I screwed up the initial packaging) 19:24:10 mull: will you guys be doing a feature page for F17? :) 19:24:30 mull: I'm sure I can find someone nice to help you :) 19:24:39 I sure hope so. Still trying to figure out what some of the hurdles will be 19:25:11 mull: well, if we can get some of hte hurdles listed, at least they're known and might get help :) 19:25:18 GWT requiring _old_ eclipse compiler is a problem... our dependence on a very old quartz version, introduction of JDK7 19:26:00 I'm not even sure what GWT is, but I'll take your word for it. :) 19:26:11 google web toolkit 19:26:16 mull: do you have most of this stuff spelled out in the Euca wiki page? 19:26:18 * rbergeron looks 19:26:37 if you've seen spot's chromium rants, you'll understand what I'm dealing with 19:26:43 right 19:26:49 Are you keeping track of fail points too? :) 19:26:54 LOL 19:27:08 more activemq, lol 19:27:47 anyway, what do I have to do for the official feature page, rbergeron ? 19:28:46 Well. I'm SO GLAD YOU ASKED 19:28:48 Let's see here. 19:29:11 #info Feature Policy! 19:29:14 * jsmith suggests http://fedoraproject.org/wiki/Features/Policy/Process as an easy way to visualize it 19:29:16 #link http://fedoraproject.org/wiki/Features/Policy/Process 19:29:24 is probably a good place to start. 19:29:29 (I will of course help you if need be.) 19:29:37 (Or you can get that gregdek guy to pitch in.) 19:29:43 (I actually recommend that. LOL) 19:29:43 haha 19:29:55 Anyway: You'll need to make a feature page. Which would look like.... 19:30:07 Haha indeed. 19:30:22 Ooh, a chart! 19:30:24 https://fedoraproject.org/wiki/Features/OpenStack 19:30:31 But, you know, with Eucalyptus. 19:30:46 * rbergeron notes that she is honest to god wearing her new eucalyptus shirt right now too 19:31:07 who needs euca when you have openstack? 19:31:12 rbergeron, which one? the blue one, or the Elastic Company one? 19:31:13 * jdarcy thinks a eucalyptus shirt would be scratchy. 19:31:20 mull: the blue one. 19:31:25 I have the elastic company one too, though 19:31:31 good. :) 19:31:40 johnmark: don't make me swat you 19:31:43 :) 19:31:49 lol 19:31:53 anyway. 19:32:04 mull: seroiusly, feel free to give a shout out on list if it's confusing. Because it is to a lot of people. 19:32:16 And it's a big project to get loaded into Fedora, but I think it's good for everyone involved. 19:32:39 And you know my opinion is *so* important... oh, wait... well, I'm sure others have hte same opinion. :) 19:32:55 #action mull to ask for help with Feature Page for Eucalyptus if he's feeling lost. 19:33:02 * rbergeron looks at gregdek 19:33:28 * gregdek reads back. 19:33:33 Sorry, running another meeting. 19:33:35 #topic Anything else? 19:33:47 johnmark, jdarcy: maybe you could tell us what changes to expect in Gluster and Heka FS land this time around? 19:33:48 Oh, yeah, I can help mull with that. 19:34:05 rbergeron: errr 19:34:10 * jsmith is still looking for volunteers to help write the Cloud Guide -- thanks to gholms for filling in some of the EC2 information 19:34:18 Heka merges into Gluster! 19:34:27 kkeithley: exactomundo 19:34:43 johnmark: and I hear there may be some 32-bit stuff happening? or am i off-track there? 19:35:03 I think there'll be a pretty extensive list of enhancements, but I'm not sure how much I'm supposed to say. 19:35:41 .........okay 19:35:52 * rbergeron wonders if she can look on a mailing list somewhere instead, har har 19:35:59 ANYWAY 19:36:07 johnmark: are you sending more people this way to learn to become superpackagers? 19:36:14 or are you going to become a packager yourself? 19:36:25 I'm working my way there. 19:36:29 I mean, if you can get ovirt running on your laptop, I have to assume you can do packaging :) 19:36:33 joejulian: hey hey :) 19:36:57 Actually that's something that I think we can talk about. Thanks to Joe, Kaleb, and others, we might have only one filesystem hierarchy for GlusterFS instead of one from Gluster and one from Fedora. 19:37:29 yay 19:37:37 gah 19:38:15 jdarcy: that will be huge 19:38:24 jdarcy: I can assume I'll probably see a feature page related to that? 19:39:02 rbergeron: Errr.... 19:39:24 okay, i'll try johnmark: can i assume that i'll see a feature page related to that? 19:39:34 sure! :) 19:39:41 not because it's marketing beauty, but because... you know... changes in filesystem hierarchy? 19:39:44 or something? 19:40:18 okay, well, we'll continue exploring as you guys figure out more about your new marriage and all that. 19:40:23 rbergeron: well, it's about folding in heka... 19:40:24 lol 19:40:44 :) 19:40:51 Ummmmmm. 19:40:56 jdarcy: oh wait, you're talking about having a single build source and spec file 19:40:58 Should we do a feature page about enhancements to GlusterFS (e.g. UFO, Hadoop)? It's not a *new* package, but I could see that. 19:40:59 Okay, other topics 19:41:19 jdarcy: hell yes 19:41:47 johnmark: I think jdarcy was talking about /usr/sbin vs /usr/local/sbin vs /opt/bin 19:42:12 kkeithley: oh right. so that different RPMs dont' clobber each other 19:42:15 That's FHS (File Hierarchy Standard or something) 19:42:25 right... which lots of people ignore :) 19:42:33 but we need to solve the single build source, spec file, etc. too 19:42:37 wait, /opt/bin?? 19:42:49 kkeithley: yes, exactly 19:43:04 rbergeron: the builds from gluster.org have always put stuff in /opt/ 19:43:16 whereas fedora builds have put things where they're supposed to go :) 19:43:19 right 19:43:33 and now we want all builds to put things where they're supposed to go 19:43:48 johnmark: and then we wwill have world peace :) 19:43:53 and eventually, we'd like to use the same .spec file 19:43:59 rbergeron: right. or something :) 19:44:33 The fedora spec just got a lot closer to what gluster uses, or at least what JoeJulian is proposing that gluster use 19:44:34 anyway, we're continuing to discuss that. not need to hash out all the details here 19:44:39 yup 19:44:43 JoeJulian: excellent 19:44:52 Sorry, X locked up as I was installing fonts.... Yes, I've redone the spec.in and it's been merged into the tree to comply with FHS. 19:45:07 JoeJulian: break out the champagne! 19:45:08 johnmark, kkeithley: excellent. joejulian: :) rock oh, you guys 19:45:19 and yes, no need to hash it out here, though we're happy to listen to people gloat about success. :) 19:46:02 anyhoo! /me shall move on 19:46:24 * rbergeron wonders if the gluster docs should go in the storage guide or cloud guide... or if we can cross post to both somehow 19:46:35 to paraphrase one of my fave PotC quotes: the FHS is more what you'd call "guidelines" than actual rules. Welcome aboard the Black Pearl, ... 19:46:37 jsmith: are we stlil doing the storage guide? 19:46:43 or did we stop that a few versions ago 19:46:50 rbergeron: Hmmmmn... I'm not sure. 19:46:56 okay 19:47:02 i'll poke ric and steve levine aobut it 19:47:03 kkeithley: LOL! I'll have to remember that 19:47:14 rbergeron: there's a storage guide? 19:47:19 johnmark: plz hold 19:47:40 I'd be happy to reuse what we have 19:47:49 http://docs.fedoraproject.org/en-US/Fedora/14/html/Storage_Administration_Guide/index.html 19:47:53 Looks like the last version of the "Storage Administration Guide" was F14 19:48:19 I'm happy to help coordinate efforts with the docs team 19:48:30 jsmith: I think someone is now like, maintaining it internally 19:48:32 but not externally 19:48:42 you might ask ric or steve levine if you want to follow up on it :) 19:48:56 steve levine does the cluster stuff, less of the filesystem stuff 19:49:01 but I think he knows who does 19:49:10 OK, will do 19:49:20 but I'm sure ric would be happy to see it getting maintained more in the community 19:49:24 jsmith: copy johnmark on it as well 19:49:29 since he's communityman in the storage BU 19:49:31 rbergeron: Will do 19:49:32 jsmith: yes, plz 19:50:58 okay. 19:51:02 ANYONE ELSE? 19:51:07 * rbergeron has a bus ready and waiting for you 19:51:24 are we done? 19:51:34 johnmark: in 10 minutes, or sooner if nobody has anything to say 19:51:40 I can just troll you for 10 minutes if you'd like 19:51:57 :D 19:52:51 BTW, I invite everyone to read my "Translator 101" series on hekafs.org and start writing their own filesystem plugins. :) 19:52:53 i would first start with the open cloud initiative stuff but you guys are back in business now :) 19:53:04 rbergeron: oy! 19:53:11 jdarcy: yes, I saw that, very cool stuff - /methinks a mail to the cloud sig list would be cool about those posts. 19:53:13 oh, hey, since this is the cloud sig 19:53:20 jdarcy: are those getting cross-posted to planet.fp.o? 19:53:25 johnmark: IT IS? 19:53:34 rbergeron: AFAIK they're being syndicated both there and on gluster.org 19:53:40 jdarcy: that is a good idea to send them to the list 19:53:55 jdarcy: yes, I syndicate everything from you on glusterfs blog now 19:54:14 johnmark: you could just add the glusterfs feed to the planet.fp.o feed 19:54:23 rbergeron: good idea 19:54:24 and see if anyone screams 19:54:27 :) 19:54:34 * jdarcy 's head asplodes. 19:54:36 YES! ANOTHER ONE! 19:54:51 screaming like that 19:54:57 jdarcy: that's bad for ur brain 19:55:04 and messy. 19:55:26 so anyhoo 19:55:43 mark your calendars for Open Cloud Conf 19:55:55 Feb. 21 and 22 19:56:05 Is that different from cloudconnect? 19:56:08 hey gregdek - you should get in on this 19:56:10 yes 19:56:28 johnmark: do i need to drop the obvious "perhaps this would be a good thing to talk about on the list as well" hint? 19:56:40 (you have 4 more minutes of being trolled, FYI) 19:56:41 rbergeron: no, no need :) 19:56:45 Going on our event calendar. 19:56:52 (actually, i should stop, since it's good to have you here) 19:56:55 gregdek: w00t 19:57:17 jdarcy: kkeithley - would be good for you guys to come out and head up some workshops 19:57:58 okay 19:58:03 johnmark: that's the days after the Brno Dev Conference :) 19:58:10 just enough time to travel 19:58:11 what kind of workshops? 19:58:14 rbergeron: yes. Hello, jet lag 19:58:22 Just need travel budget 19:58:28 kkeithley: "this is how you utilize GlusterFS" workshops 19:58:36 as well as "this is how you build translators" 19:58:42 very hands on 19:58:48 kkeithley: I know whom to ask about that :) 19:59:01 johnmark: CAN YOU MAKE THAT HAPPEN? 19:59:07 hahaha 19:59:10 johnmark: Did you give a URL for the conference? Google's failing me. 19:59:12 rbergeron: stop :) 19:59:19 we just purchased the URL today 19:59:25 and by "we" I mean me and Dave Nielsen 19:59:25 johnmark: i can't help it ;) 19:59:32 opencloudconf.com 19:59:33 lol 19:59:59 Ah, OK. I'll keep an eye on it. 20:00:04 cool 20:00:06 johnmark: is that an OCI-related/endorsed thing? Or not so much? 20:00:35 rbergeron: it will be, yes 20:00:47 coolio 20:00:53 okay, ummmmm. 20:01:14 It's beer o'clock. 20:01:18 No kidding. 20:01:46 Anyone else have any exciting things to talk about? 20:01:49 :D 20:02:03 if not, i'll strap on my ianweller-approved jetpack and light the fuse. 20:02:12 * gholms takes cover 20:02:20 gholms: i didn't actually build it myself 20:02:41 * rbergeron counts backwards from 87 20:02:46 56 20:02:49 28 20:02:59 sweet, sweet 16 (as in fedora 16) 20:03:02 .... 5 20:03:03 4 20:03:05 ...1 20:03:10 Thanks for coming, folks. 20:03:13 err. 20:03:14 PEEPS. 20:03:16 #endmeeting