21:02:11 #startmeeting 21:02:12 Meeting started Thu Feb 11 21:02:11 2010 UTC. The chair is gregdek-mobile. Information about MeetBot at http://wiki.debian.org/MeetBot. 21:02:13 Useful Commands: #action #agreed #halp #info #idea #link #topic. 21:02:22 w00t 21:02:39 * gholms points gregdek-mobile to http://fpaste.org/mkI7/ 21:02:51 ok, any objection to starting with todos? 21:03:13 * gregdek-mobile needs someone else to post them one by one for discussion :) 21:03:29 * EC2 Spin Status 21:03:41 * gholms can't use #topic without being #chair'ed 21:03:52 #topic covering todos from last meeting 21:04:01 somebody #chair gholms and jforbes 21:04:15 s/somebody/gregdek-mobile 21:04:37 but it says I'm chair! 21:04:46 grrr @ meetbot 21:04:51 gregdek-mobile: It can be a list. :) 21:05:22 anyway. 21:05:45 jforbes: how's the latest kernel? any movement? 21:06:40 gregdek-mobile: lots of movement, but no ETA. Basically there are issues with .32 (granted, there are issues with .31, but it is already there, and people don't seem to like new issues even if old issues are fixed) 21:07:01 make sure to use liberal amounts of #info 21:07:02 gregdek-mobile: issues in this context, not referring to EC2 21:07:06 for the minutes 21:07:25 * gholms recommends finer-grained #topics than "todos from last meeting" 21:07:26 #info some progress on .32 kernel but no eta 21:08:08 gholms, can I chair you? 21:08:17 my typing is suboptimal at present 21:08:31 #chair gholms 21:08:32 Current chairs: gholms gregdek-mobile 21:08:33 gregdek-mobile: Sure; use the #chair command 21:08:39 #chair jforbes 21:08:40 Current chairs: gholms gregdek-mobile jforbes 21:08:52 ok. 21:09:13 Cool 21:09:22 #topic EC2 Kernel 21:09:58 seems like we're done with kernel.... hurry up and wait, yes? 21:10:06 Any word on documenting the aki/ari upload procesS? 21:10:13 correct 21:10:24 gholms: no, It will be documented as I do the first upload 21:10:25 s/S/s/ 21:10:47 Well, technically I have some documentation that just needs to be modified for Fedora 21:11:51 I presume upload credentials for kernel maintainers are contingent on that. 21:11:58 I suppose at some point we need to discuss bucket labeling 21:12:06 gholms: Yeah, contingent on that 21:12:14 I think 2 buckets is sufficient. 21:12:14 gholms: but not holding us back at all 21:12:18 and we have both. 21:12:37 Anything else kernel-related? Next up is update mirroring. 21:12:41 1 for authoritative, and 1 for all other work including test kernels. 21:12:51 gregdek-mobile: That's accounts 21:13:23 oh... right. 21:13:23 gregdek-mobile: I am thinking about storage buckets for the official stuff, I am guessing F#-arch would make the most sense 21:13:40 yeah, +1 21:13:55 Kernels and amis in the same buckets? 21:14:08 gholms: might as well be 21:14:13 Makes sense to me. 21:15:02 #info kernels and amis in same bucket per version 21:15:28 Something like f12-i386 and f12-i386-testing? 21:15:54 gholms: testing won't appear in the official account 21:16:07 #info kernels and amis in same version per version-arch 21:16:30 dammit, I'll get it right in the wiki notes :) 21:16:57 gregdek-mobile: If you want to try typing it again you can use #undo ;) 21:17:04 heh, nah 21:17:15 Anything else kernel-related? 21:17:16 I'm going slow enough as is. :) 21:17:21 That's it for kernel 21:17:29 #topic EC2 Update/Source Mirrors 21:17:37 * gholms hands gregdek-mobile a keyboard 21:17:46 this is gonna take a while. 21:17:55 gregdek-mobile: How are things going with getting aki hosting on S3? 21:17:59 waiting on a meeting with nathan thomas to discuss. 21:18:12 goal is to get amazon to sponsor a mirror. 21:18:32 fallback is to pay, but we would need an estimate. 21:18:39 #info goal is to get Amazon to sponsor an aki source mirror 21:18:49 if too high, that will be unfeasible. 21:19:13 so waiting on meeting, basically. 21:19:48 For update mirrors we will need one not just in every region, but in every availability zone within each region. 21:20:01 yep. 21:20:09 it's a fair few. 21:20:17 but we're built for it. 21:20:29 Do we know if Amazon has set IP ranges for each availability zone so they can get into MM? 21:21:00 I assume so... but we should def verify. 21:21:31 #action gregdek will verify that amazon can provide ip ranges for availability zones for mirrormanager 21:22:07 We might also want to figure out how this works with hosts that use elastic addressing. 21:22:41 I'm guessing that mm would handle that? 21:22:47 * Oxf13 steps out 21:22:58 FWIW, they have done this with Ubuntu, so should know what is required for parity 21:22:59 rather, yum would be able to figure it out? 21:23:06 yeah. 21:23:41 that's sort of the goal: say "what you did w ubuntu, can we do the same?" 21:24:15 I think the hard part is going to be making them not only hit the correct mirrors for the zones that they are in, but also via their internal addresses instead of their external ones. 21:24:30 Just something to ponder. 21:24:30 ooooo yeah maybe. 21:24:44 yeah, we'll get it sorted. 21:24:50 what's next? 21:25:01 #topic euca2ools and eucalyptus packages 21:25:26 yeah, how's that going? any roadblocks? 21:25:41 euca2ools is now in updates-testing. Everyone try it out and comment in bodhi. ;) 21:26:08 woooo! 21:26:25 we should advertise that. 21:26:32 s/should/will 21:26:55 #action gregdek advertise availability of euca2ools in bodhi 21:26:58 There's a bug with python-boto-1.9b-1 that breaks euca-register. It should be fixed with the next python-boto release. 21:27:09 ew, ok. 21:27:34 (Meaning the bug was CLOSED NEXTRELEASE) 21:27:45 how bad? does it require amazon's tools to register for now? 21:27:58 It's completely non-functional right now. 21:28:18 maybe we should wait for that advertisement. 21:28:28 what does python-boto do? 21:28:29 http://koji.fedoraproject.org/koji/buildinfo?buildID=155030 21:28:41 That's the backend for EC2 and S3 that euca2ools calls into. 21:28:52 ah. 21:29:04 so beyond our control completely then. 21:29:10 internal to amazon? 21:29:46 Oh, the new python-boto seems to have hit updates-testing. So we should be good. :) 21:29:55 yay! 21:30:04 all right then. 21:30:09 * gholms stops putting his foot in his mouth 21:30:29 Packaging eucalyptus itself is going to be a bit of work. 21:31:08 right... there was some interest on cloud list. any followup? like from mattdm for example? 21:32:36 There have been some comments on the list that will hopefully push upstream's spec file toward meeting packaging guidelines. 21:33:16 anything we can/should do to move that along? 21:34:46 can I presume that silence is a no? :) 21:34:50 Graziano from upstream seems to have been very receptive to my mini package review. 21:35:02 oh, that's good. 21:35:07 Maybe giving that process some more time will help get us close. 21:35:22 but I guess we're just waiting on them for their next submission? 21:35:53 Looks like it. 21:35:59 I can't test eucalyptus, so I'm hesitant to head up a packaging effort for it. 21:36:28 mmm hmm. 21:37:05 well, let's focus on getting euca2ools past the finish line and we'll keep an eye on eucalyptus for now, 21:37:46 anything to add re: euca-* pkging? 21:37:46 #info euca2ools is in updates-testing for F12 21:38:19 #info Discussion on cloud list to improve eucalyptus's spec file's compliance with packaging guidelines 21:38:27 Anything else, anyone? 21:38:37 not here 21:39:03 * gregdek-mobile thinks a lot of folks are satisfied to lurk this time :) 21:39:07 #topic IBM Cloud 21:39:57 * gregdek-mobile doesn't even remember this one. 21:40:13 whose todo is it? probably mine. lol 21:40:13 Unfortunately the only other person with input to the IBM cloud is gone. 21:40:19 yo 21:40:23 :D 21:40:33 gregdek-mobile: it came up on the list, Oxf13 was one of the people looking into it 21:40:35 jforbes... who is that? 21:41:04 * gregdek-mobile clearly has not been keeping up with email well in the last couple of days... 21:41:19 gregdek-mobile: jkeating 21:41:20 what's the gist? 21:41:32 never mind... I'll read up. 21:41:40 anything actionable here? 21:42:11 gregdek-mobile: at this point they are looking to see what is required to have a presence there 21:42:36 There was someone on the list who plans to meet with IBM people about it. Make sure to send him lots of questions. 21:42:41 ok. prelim investigation sort of stuff? do we have ibm contacts? 21:42:45 ok. 21:43:23 * gholms doesn't know Oliver's nick 21:43:30 #action gregdek will meddle, likely unhelpfully, in ibm conversation 21:43:40 Yeah, prelims, unknown if we have contacts there 21:43:56 meeting still going on? 21:44:01 FWIW, I've been working (and failing) to get Fedora up and running on IBM's cloud. 21:44:02 yep :) 21:44:18 just going thru open work items 21:44:35 anything else to say about ibm? sounds like not... 21:44:36 * gholms notices something missing from /topic 21:44:53 hrm? 21:44:55 #meetingname Fedora Cloud SIG (2010-02-11) 21:44:55 The meeting name has been set to 'fedora_cloud_sig_(2010-02-11)' 21:45:11 ah, heh, thx 21:45:20 Dang, looks like you can only set that when you #startmeeting 21:45:33 Anything else IBM-related? 21:45:46 * jsmith lurks 21:45:52 what is the cloud 21:45:53 so why does everyone show up now? :) 21:45:58 SHUT IT 21:46:03 what is the IBM cloud and its interatction with you 21:46:03 Because it's time for... 21:46:08 gregdek-mobile: someone put a note in #fedora-admin a bit ago :) 21:46:36 lemme guess: "plz troll the cloud meeting!" :) 21:46:48 I asked mmcgrath to show up. ;) 21:46:50 smooge, is that a serious q? 21:46:57 gregdek-mobile, sorry I hit return to quickly wasn't meaning a 'joke' more like what do you need with IBM cloud as I missed this discussion 21:47:06 am trying to catch up 21:47:09 because our answer so far is "we dunno but we're gonna find out", it seems. 21:47:14 gregdek-mobile, we were talking about this yesterday in -admin 21:47:15 ah ok. 21:47:23 several of us got new accounts and instances running 21:47:29 * jsmith has VMs running 21:47:30 can someone post meetbot log url for smooge? 21:47:46 mdomsch, in ibm cloud? 21:47:48 yes 21:47:55 shiny! 21:47:55 http://meetbot.fedoraproject.org/fedora-cloud/2010-02-11/fedora_cloud_sig_%282010-02-11%29.2010-02-11-21.02.log.txt 21:47:57 it came up in the brian stevens webinar yesterday 21:48:04 gholms, thanks 21:48:05 who set it up? 21:48:05 so I jumped on, then mmcgrath, then jwb, ... 21:48:27 gregdek-mobile, the next question was how could infrastructure help if any 21:49:01 it looks like they're running KVM on RHEL5.4, and are heavily over-subscribed ATM 21:49:08 "victim of our own success" she called it 21:49:19 Haha 21:49:28 lol 21:49:37 I've had about 8 guests up and running 21:49:51 so how does one get accounts on the ibm cloud? 21:49:58 fill out the form at ibm.com/cloud 21:50:07 trivial, and free (for now) 21:50:15 nice! 21:50:27 took me about 30 minutes from figuring out it existed (thanks mdomsch) to me logging into a guest 21:50:36 jwb laughed - we said it was so easy to do - he thought that odd coming from IBM 21:50:39 #link http://www.ibm.com/ibm/cloud/ 21:50:43 what flavors? fedora 12? 21:50:49 RHEL 5.4, SLES 11 21:50:51 no fedora 21:50:58 hm. 21:50:59 gregdek-mobile: I don't see any unsupported OSes at all actually 21:50:59 no RHEL repos besides base 21:51:01 …yet. :) 21:51:10 right. :) 21:51:12 You can add repos to running instances, though, right? 21:51:23 so someone needs to investigate a lot of that stuff. 21:51:26 it's aimed at users of their WebSphere, DB2, and Rational development tools 21:51:34 gholms: you can, but I tried to rhn_register and got failboat 21:51:43 :-\ 21:51:54 ew. 21:52:10 But you can rhn_register when you don't have any third-party repos? 21:52:10 and the instances aren't persistent with disk, unless you create another volume to mount at /somewhere 21:52:22 mdomsch: actually that's not quite true. 21:52:31 I've persisted across reboots 21:52:36 oh? 21:52:38 I don't know what that means exactly. 21:52:47 hunh. 21:52:48 yep 21:52:51 touch /test 21:52:52 reboot 21:52:54 it was still there 21:52:55 that's amazon EC2-like 21:53:09 only if you're using ebs, I thought. 21:53:12 mmcgrath, interesting. 21:54:23 anyway... since there's no fedora action there other than "see if we can even", shall we move on? 21:54:41 yup yup 21:54:44 they claim to have APIs you can use to start a new image from scratch (boot from ISO), which is where I'd start for a Fedora image 21:55:07 mdomsch, oooh, we should look at that 21:55:23 who can look for those mystery apis? 21:55:27 * mmcgrath notes they block icmp pings which is annoying 21:55:31 lol 21:55:37 * gholms rolls eyes 21:55:41 ping is 4 punks 21:56:32 Anything else for this topic? Anyone able to look into their APIs and processes? 21:56:59 id like to assign an action, but not if no one can take it 21:57:10 gregdek-mobile: APIs/HowTo-Wiki for IBM cloud https://www.ibm.com/developerworks/mydeveloperworks/wikis/home/wiki/Wa3175e7ea2f3_402b_9a95_c5e32ae46ac5?lang=en_US 21:57:20 and API http://www-949.ibm.com/cloud/developer/ram/assetDetail/generalDetails.faces?guid={AE9C1E74-F8C5-A11C-DDE3-D0E6F12515FE} 21:57:31 jmh can you #info those? 21:57:38 or #link? 21:57:51 Doesn't it pick up links on its own? 21:58:04 oh yes, I guess it does. 21:58:10 Moving on... 21:58:11 #topic Open Floor 21:58:21 * gregdek-mobile is not deeply versed in meetbot-fu. :) 21:58:48 anyone? anyone? bueller? 21:59:27 * gholms will close in a few then 21:59:53 worksforme 22:00:52 Thanks for coming, everyone. 22:00:59 #endmeeting