19:00:01 #startmeeting Infrastructure (2013-01-03) 19:00:01 Meeting started Thu Jan 3 19:00:01 2013 UTC. The chair is nirik. Information about MeetBot at http://wiki.debian.org/MeetBot. 19:00:01 Useful Commands: #action #agreed #halp #info #idea #link #topic. 19:00:01 #meetingname infrastructure 19:00:01 The meeting name has been set to 'infrastructure' 19:00:02 #topic welcome to the new bactun 19:00:02 #chair smooge skvidal CodeBlock ricky nirik abadger1999 lmacken dgilmore mdomsch threebean 19:00:02 Current chairs: CodeBlock abadger1999 dgilmore lmacken mdomsch nirik ricky skvidal smooge threebean 19:00:13 hi 19:00:13 who's around for a nice infra meeting? :) 19:00:21 * skvidal is 19:00:21 hola 19:00:26 * puiterwijk is 19:00:31 here! 19:00:40 <_xtrashot_> hello 19:00:48 fchiulli is listening in 19:01:06 * lmacken 19:01:11 * threebean threebean threebean 19:01:41 #topic New folks introductions and Apprentice tasks 19:01:41 threebean: shouldn't that just be beanbeanbean? 19:01:45 ha. 19:01:48 point taken 19:02:03 * relrod is here 19:02:03 any new folks like to introduce themselves? or have questions/comments? 19:02:06 skvidal: some irc networks have name character limits. sometimes it's nice to pick a shorter version. 19:02:12 (for consistency) 19:02:24 <_xtrashot_> Hi all! I'm xtrashot ( or alex ), I come from a solaris background, I run my own linux server @home and I would like to lend a hand to fedora on the sysadmin side of the things. 19:02:25 it could also be 'bean^3' ? 19:02:36 _xtrashot_: welcome! 19:02:53 <_xtrashot_> thanks nirik! 19:03:01 _xtrashot_: see me after the meeting over in #fedora-admin and we can get you added to our apprentice group and point you to things to look at. 19:03:04 Hi. My name is Rob and I am a Linux admin here trying to lend a hand where possible too 19:03:10 hi 19:03:13 welcome rdesanno also. :) 19:04:02 and Adran. 19:04:10 Hi nirik! 19:04:32 happy to help point you all to resources after the meeting... and do feel free to chime in with questions in the meeting here, or #fedora-admin at any point. 19:04:41 <_xtrashot_> thanks. will do 19:04:46 Will do and thanks! Looking forward to helping 19:04:49 Will do! 19:05:02 great. Glad to see new folks. ;) 19:05:24 mixing things up a bit... lets mention the freeze first: 19:05:26 #topic Freeze reminder 19:05:37 we are still in final freeze until f18 goes out the door. 19:05:47 * skvidal has noticed - it's REALLY cold in here! 19:05:53 there's a go/no-go meeting in #fedora-meeting-1, but it looks like it will be no-go. ;) 19:06:09 well I expected as such since there isn't an RC yet 19:06:10 alright! fedora 18 during fudcon? 19:06:16 or maybe after? 19:06:25 just before probibly... 19:06:34 I am voting for 2018 19:06:35 or we might go for just a few days slip this time. 19:06:45 we can then do once per year releases afterwords 19:06:51 anyhow, the upshot is that we are frozen, don't make any bug changes without approvals, etc. 19:07:12 I also have filed out release tickets... 19:07:29 most of which need something to stage before we can do anything. :) 19:07:53 #topic Applications status / discussion 19:08:04 ok, any applications status or things to note? 19:08:17 given the holidays and freeze, I suspect not much... 19:08:29 do we have anything we want to look at setting up to demo at fudcon? 19:08:34 yes 19:08:46 I have 3 talks in mind for fudcon 19:09:01 1. private cloud 19:09:02 2. ansible 19:09:04 3. copr 19:09:43 #topic Fudcon 19:09:50 might as well go to the fudcon topic. ;) 19:09:53 cool :) 19:09:55 19:10:07 https://fedoraproject.org/wiki/Fedora_infrastructure_tasks_2013 19:10:16 is an ideas container/setup for our fudcon plans. 19:10:26 I added some more things there too. 19:10:39 how do we want to try and organize things? 19:10:53 I suspect to meet during one of the hackfest times? 19:10:57 and in the evenings 19:11:19 I'm fine with a generic Infrastructure hackfest and try and cover a bunch of things that need high BW, but might be nice to have a checklist of things we should try and address. 19:11:54 I see the bits you've added 19:12:06 some of those have been on the desired-list for quite a while 19:12:10 yeah. 19:12:27 never enough hours in the day. 19:12:53 indeed 19:13:00 and some of them are not overly difficult 19:13:08 but they involve a bunch of people and moving parts 19:13:16 so the inertia is high (fedorahosted as an example) 19:13:23 something maybe for fudcon 19:13:23 yep. 19:13:26 maybe for before 19:13:39 is to go through all the [cr]apps we're supporting 19:13:43 and get rid of those unused 19:13:56 we killed off the old fedora community before the break 19:13:58 and that was great 19:14:05 id like to talk about replacing the ca infrastructure 19:14:05 but I have an item on my list to ask about insight now 19:14:22 dgilmore: what ca infrastructure? it's just easy-pki :) 19:14:24 so, the general fudcon setup is barcamp first day... then hackfests next two 19:14:35 skvidal: what we use for koji and users certs 19:14:47 dgilmore: ah, gotcha 19:14:51 skvidal: its some make files 19:14:54 nod 19:14:58 and a python script 19:15:37 dgilmore: what would you like to change it to? 19:15:52 nirik: im thinking dogtag 19:16:16 that seems like a lot of overhead... 19:16:16 dgilmore: what's the selling point? 19:16:26 though I admit I haven't looked at it recently 19:16:54 skvidal: that we could have two nodes, and fas can talk to it over the network, so all fas boxes could handle cert requests, 19:17:04 eliminate spof 19:17:37 we would also not have a static crl and have real time checking on revoked certs 19:17:49 could nfs/gluster mount also fix that? 19:18:11 nirik: not without major locking checking 19:18:25 and it still doesnt give us real time checking 19:18:39 well, we can discuss it. dogtag has always seemed very heavy to me... 19:18:41 not that its a huge deal to fetch the crl every hour 19:18:55 nirik: maybe its a bit too heavy for us 19:19:23 we can see what makes sense... 19:19:45 nirik: skvidal: id like a config management talk also 19:19:54 so we can go over things we have changed 19:19:58 dgilmore: that's the ansible talk 19:20:06 yeah... 19:20:07 id like to get arm setup exactly the same as primary 19:20:20 and I think it's less about what's changeD and more about what is still changing 19:20:29 but the point is still the same 19:20:31 skvidal: sure 19:20:52 * nirik wonders if we should try and have a generic '2013 planning for infra' barcamp session... but that sounds boring. Perhaps we just try and do that at the start of hackfests. 19:21:41 anyone else looking at doing barcamp sessions? 19:21:59 Hmm.. I'd like to do a fedmsg talk; how should I aim it? More towards what's been done or more towards where things should go? 19:22:01 fedmsg session! 19:22:12 both 19:22:15 yeah... 19:22:27 go over whats been done and talk some about upcoming? 19:22:33 cool, sure. 19:23:03 lmacken: any chance we could stand up a mockup of bodhi 2.0 for people to look at? or still too early? 19:23:54 nirik: might be too early for that, but hopefully a bodhi2 hackfest will happen 19:24:09 ok. 19:25:03 * nirik might pitch a 'make rawhide nicer to use' session, but thats not really much infra related. ;) 19:25:28 nirik: maybe i could run a using enterprise arm hardware session 19:25:35 I'd also like us to see if we can decide a fad topic... kock something out in 2013 thats difficult to get done remote. 19:25:37 not sure what the iterest would be 19:26:18 dgilmore: yeah. I bet there will be a few arm sessions... it would be cool if we could show off some server arm stuff (if even only remote) 19:26:30 nirik: we expect to have an arm server on site 19:27:00 bconoboy: sweet. ;) 19:27:09 arm barcamp? 19:27:18 barmcamp? 19:27:20 no sorry 19:27:23 :) 19:27:43 ok, so, any folks doing barcamp sessions, please do go add them to the fudcon page now... 19:27:49 nirik: hopefully we can show some stuff off 19:27:55 * bconoboy does so 19:28:21 I will add a infra hackfest for sat/sun and we can use the first part of sat to try and plan out what we want to try and get done and pull in anyone we need for those things. 19:28:31 unless we can plan that in advance? 19:28:59 * nirik thinks it might be better higher BW 19:29:11 nod 19:29:26 I'll try and organize a bit more if I can... 19:29:31 theoreticaly we should have everyone nearby-ish to put a bullet in insight* 19:29:46 do we have anything else on our 'should die' list? 19:30:05 yeah, do we want to mail people now about that? I guess lists and asrob and stickster? 19:30:10 * skvidal hmms - maybe I should check last login on all systems by a user other than root 19:31:37 I don't have anything on that list off hand... 19:32:02 any other fudcon business? 19:32:30 #topic Upcoming items 19:32:34 #topic Upcoming Tasks/Items 19:32:34 #info 2012-12-11 to 2013-01-08 F18 Final Freeze 19:32:34 #info 2012-01-05 PHX2 outage. 19:32:34 #info 2013-01-08 F18 release. 19:32:35 #info 2013-01-18 to 2013-01-20 FUDCON Lawrence 19:32:36 #info 2013-01-29 f19 feature submission deadline. 19:32:46 we have a phx2 network outage this weekend. 19:32:53 it's supposed to be very short, but we will see. 19:33:02 I will be around for it 19:33:10 * relrod has a note in his calendar to nuke proxy01 from dns a few hours ahead of time. 19:33:14 * nirik will be too. 19:33:23 #topic Open Floor 19:33:31 any general topics or questions? 19:34:03 so 19:34:07 about our web apps/etc 19:34:27 I was talking with nirik yesterday about thinking of ways of making our apps easier to debug when things go wrong 19:34:49 as it is right now we end up trying to figure out if an app is misbehaving, being hit too much from the outside or if it is being impacted by SOMETHING ELSE 19:35:04 and it means a lot of our fixes are somewhat imprecise (restart httpd) 19:35:25 I was debating some ways of dealing with that - it's a subject I suggested to discuss at fudcon 19:35:27 I think moving to a model of 'one app per virtual' might be good for that. 19:35:35 I like that one 19:35:37 but I wanted to toss it out here to see if anyone wanted to scream murder 19:35:41 but containers, etc might also work, not sure. 19:36:12 I remember we wanted to look at that back a couple of fudcons ago but ran into "how will we manage a wave of them" 19:36:25 I think we know how to manage them 19:36:27 and now we have the answer.. nirikseth cloud 19:36:37 not murder.. but.. the same problem might be solved just as well by fixing our logging setup instead. 19:36:37 and ansible 19:36:45 right.... so.. 19:36:53 a consideration, definitely. 19:37:50 threebean, well one of the issues we run into is that we have 8 different apps on one box.. and it has turned into that A impacts B but B complains out because it is a delicate flower of an app. then C and D get into a fight and E goes on marxist strike. 19:38:11 well, A OOMs apache, and BCDE all also die. 19:38:33 * threebean nods 19:38:49 More separation is better, then. :) 19:39:13 part of this issue is the roles our smaller sites play. 19:39:17 then I figure logging will help on the A needs B but B doesn't want to play with A without C 19:39:52 I think on the new apps we have deployed to their own app servers it's worked out ok... packages, ask, paste... 19:40:02 but some of the 'core' ones might be different 19:41:11 anyway - I just wanted to float it 19:41:21 it also might have the advantage of letting us get to know the apps a bit better 19:41:24 well it's the same thing with the db servers, many app dbs per server -- we started solving that by moving the fas one separate of the others, I guess 19:41:36 since we'd spend less time learning how the apps interact w/each other 19:41:39 and more how they act on their own 19:41:57 relrod: agreed - but same thing -we've moved some apps with special db needs to their own isolation 19:42:17 relrod: fas was just moved to keep from having a looping SPOF in the shape of the fas and db servers 19:42:33 right, and can avoid stack issues like TG1 / TG2 / pyramid in the same app server... 19:42:46 anyhow, yeah, I think it's a good plan to discuss more at fudcon 19:43:01 agreed 19:43:09 okay cool 19:43:10 thanks! 19:44:23 that's all I had 19:44:30 other than to say 19:44:31 die insight 19:44:32 die 19:44:36 unless it is being used 19:44:38 then that's great 19:45:54 ha. ;) 19:46:00 it's really not 19:46:07 anyhow... 19:46:12 anything else general of note? 19:46:28 or shall we close out and go back to our frozen existance? 19:46:29 not for me 19:46:49 yeah, good here. I'm pumped for fudcon 19:47:20 threebean: yeah, should be fun :) 19:47:50 yeah, fudcons are always great. 19:48:10 I hear they are fun. Never been my self though. :\ 19:48:10 * skvidal wonders if spot is going to have a copy of cards against humanity for boardgames 19:48:25 * nirik needs to come up with an agenda tho of what he wants to get done. otherwise it's very easy to get sidetracked into doing other things. ;) 19:48:26 skvidal: yep. plus expansion pack 1. 19:48:28 I played it over the break and.... well let's just say I'm well suited to the game 19:48:31 spot: ooooo 19:48:39 yeah, it's an awesome game. :) 19:48:45 :D 19:48:50 it's a horrible game for terrible people 19:48:56 whoa, nice. 19:48:56 like mytself 19:49:05 heh. :) 19:49:11 ok, thanks everyone... 19:49:15 #endmeeting