18:59:59 #startmeeting Infrastructure (2013-07-11) 18:59:59 Meeting started Thu Jul 11 18:59:59 2013 UTC. The chair is nirik. Information about MeetBot at http://wiki.debian.org/MeetBot. 18:59:59 Useful Commands: #action #agreed #halp #info #idea #link #topic. 18:59:59 #meetingname infrastructure 18:59:59 #topic welcome y'all 18:59:59 #chair smooge CodeBlock ricky nirik abadger1999 lmacken dgilmore mdomsch threebean 18:59:59 The meeting name has been set to 'infrastructure' 18:59:59 Current chairs: CodeBlock abadger1999 dgilmore lmacken mdomsch nirik ricky smooge threebean 19:00:29 who all is around? 19:00:30 * pingou here 19:00:33 * threebean is here 19:00:38 * rhaen is here 19:00:58 * lmacken 19:01:43 hey 19:02:13 ok, lets go ahead then... 19:02:17 here 19:02:17 * tflink is lurking 19:02:20 #topic New folks introductions and Apprentice tasks. 19:02:29 yo 19:02:30 any new folks like to say hi, or apprentices with questions? 19:02:39 Here. 19:03:04 Hello fedora team, I am a sysadmin/dev in real life 19:03:05 just hi 19:03:10 and would like to join. 19:03:13 rhaen: hi :) 19:03:17 welcome rhaen 19:03:20 * oddshocks waves at rhaen 19:03:22 welcome rhaen 19:03:23 * rhaen bows. 19:03:31 are you more interested in sysadmin or development stuff? or both? 19:03:58 nirik: actually, I haven't wrapped my mind around it. I am spending building packages for the company I am working for 19:04:05 and do lots and lots of puppet coding 19:04:17 So I think I could help with the puppet stuff. 19:04:22 rhaen: ok. we do use puppet, but we are moving to ansible. ;) 19:04:38 help with things is always welcome. ;) 19:04:43 oh, that's not an issue - 19:04:55 I am looking forward to :) 19:04:55 see me after the meeting and I can add you to our apprentice group to look around... 19:05:04 sure! Thanks, nirik 19:05:09 also... 19:05:39 #info nirik has slacked and not done his monthly apprentice group purge yet... will do it later today. If you're an existing apprentice, please reply to the status update email. :) 19:06:20 #topic moment of silence 19:06:56 as most of you know, we lost our friend and fedora family member skvidal this week. 19:07:29 please do keep him and his family and friends in your thoughts. 19:07:44 and be patient with us as we all try and move forward. 19:08:33 #topic Applications status / discussion 19:08:53 any news in the application world... 19:09:18 anyone who wants to review this pull request is invited to do so: https://github.com/fedora-infra/python-fedora/pull/45 19:09:35 it's a pretty invasive change to python-fedora's client. 19:09:39 more eyes are better. 19:10:10 * threebean nods 19:10:29 #link https://apps.stg.fedoraproject.org/badges/ 19:10:58 badges are moving ahead. we've been struggling with css recently. there's plenty of other work to be done, but there's progress. 19:11:19 One of puiterwijk's comments: https://github.com/fedora-infra/python-fedora/pull/45#discussion-diff-5007386 needs input -- I think we should be doing something about that but I'm not sure we should be doing it as part ofthis pull request/release or the next release. 19:11:53 threebean / oddshocks are we on target to give badges in prod at flock? or hard to say? 19:12:10 nirik: on track, yes. 19:13:05 abadger1999: so with this we could look at implementing 2fa in the webapps? it would require more work on each of them? or is plugin? 19:13:40 nirik: they will all require more work anyway at, least the input field (unless they use openid) 19:13:46 nirik: it is the first step for the client. 19:13:52 ok. 19:14:01 it makes it so the client has an api that can handle otps 19:14:05 (in two different ways) 19:14:18 the servers then need to be able to handle the otps. 19:14:37 might be nice to add a planning doc to the wiki with the various stages of what we are thinking... in https://fedoraproject.org/wiki/Category:Infrastructure_Planning 19:15:09 I think that really requires that we modify fas to handle them... SmootherFrOgZ had some work to do that but he might have to modify it a bit since the clientside api is different than what he originally wrote. 19:15:31 once fas is modified, we can look at modifying fas_openid and other webapps that use fas to authenticate. 19:15:42 k 19:16:30 one sidenote to mention.. 19:17:22 sources.debian.org popped up not long ago. its a flask app to search/show sources for all their packages. It might be a neat thing to have for us, and could mean just implemeting in packages? 19:17:40 shows package contents? 19:17:50 * nirik gets an example 19:17:57 threebean: packages has this but needs to be fixed 19:18:03 http://sources.debian.net/src/calibre/0.7.7%2Bdfsg-1squeeze1 19:18:19 https://apps.fedoraproject.org/packages/guake/contents 19:18:27 they also have a search function via that 19:18:34 not sure if it's flask or something else. 19:18:49 http://codesearch.debian.net/about 19:18:52 ha, it's in 'go' 19:18:59 oooooooooh 19:19:24 anyhow, not that we need more things, but this might be a nice side project or good way to get involved if someone finds it interesting. 19:19:37 * threebean nods 19:19:48 the front-end is flask indeed 19:20:20 threebean: isn't/wasn't packages searching content as well at some point? 19:21:09 I don't think so 19:21:12 #info see #fedora-apps if you are interested in working on codesearch or showing code for pkgs 19:21:21 it ought to though.. should just be a matter of adding it to xapian during the indexing cronjob. 19:21:34 +1 19:21:41 I miss-remember then 19:22:18 anyhow, any further apps news? or shall we move on? 19:22:45 nah, none here. 19:22:57 just to note: https://fedorahosted.org/fedora-infrastructure/report/15 is the list of our hotfixes currently. :) Do consider new releases before f20alpha to clean up some. 19:23:24 #topic Sysadmin status / discussion 19:23:31 so, on the sysadmin side... 19:24:07 I'm still more commited than ever to moving to ansible, but our progress will be definitely slowed as Seth was working on our tooling. ;( 19:24:33 yeah :/ 19:24:39 #info nirik to be in phx2 july 29-31st 19:25:00 #info outage for the 30th to add memory and drives to systems, etc. 19:25:14 #info koji outage next week... thursday night and friday. 19:25:20 nirik: for ansible -- are we at a place where we could have a 1-2 hour irc or google hangout training? 19:25:38 ie -- what the various folders in the ansible git repo are for. 19:25:47 walk through of creating a new host for a particular purpose. 19:26:19 maybe some common things that get changed like proxies, app servers? 19:26:20 abadger1999: possibly. The thing that was being worked on was the parts to allow non sysadmin-main to run playbooks on hosts they could sudo on, and a trigger/cronjob to run playbooks on changes. I was hoping to get those in place and then do some training... 19:26:31 19:26:50 I'd be happy to do a irc/hangout meetup tho to help get people up to speed... 19:26:58 are vagrant boxes an option for this? 19:27:00 and solicit help/ideas for the parts we still need. 19:27:11 rhaen: nope, we don't use vagrant. ;) 19:27:34 do we want to do it online or at flock? 19:27:46 (one being not entirely against the other of course) 19:27:50 rhaen: we actually have the virt setup story down pretty nicely... you can just define variables in ansible and it makes it for you, installs, and runs playbooks. 19:28:12 nirik: yep, got it - already looked into it. 19:28:31 I think vagrant is premade images? (I've not used it much tho, so not sure) 19:28:39 pingou: we could do both possibly. 19:28:40 I thought just for the training to provide anyone with a "fedora alike" environment for testing and educational stuff 19:28:57 rhaen: ah, I see, misunderstood... yeah, you can use any box you like to play with ansible. 19:29:01 vagrant also doesn't work well with other-than-virtualbox vms, IIRC 19:29:27 tflink: true, it supports vmware, too - however no kvms 19:29:36 how about I try and setup a time next week sometime? and we can at least gather and I can show what we have now and we can plan and see what our gaps are? 19:29:51 sounds good 19:29:52 nirik: sounds good. 19:30:19 nirik: for the "run playbooks on changes" trigger.. did you have a git hook in mind? 19:30:24 #info nirik to plan an ansible session next week to bring people up to speed and discuss plans. 19:31:09 threebean: yeah. We (well, Seth) was going to be more clever than that tho... and figure out what hosts were affected by the change. So, if you check in a badges01 change it would only re-run the badges playbook. 19:31:15 looks like someone doesn't like freenode again 19:31:35 nirik: yeah, I'm thinking about it now. 19:31:43 we wanted to move to a more triggered model than a run every 30min model that we have in puppet. 19:31:46 smooge: yep 19:31:56 sometimes of course you can't tell, and have to just run everything. 19:31:57 but, a smart git hook would be cool? you push, and it updates only hosts that need updating? 19:32:11 threebean, yeah 19:32:43 threebean: yeah, although we also talked about making it time delayed a bit... ie, so you could commit and someone could see your commit and tell you about a mistake before it ran... 19:32:49 but thats all up in the air. 19:33:17 I remembered that last part 19:33:22 that sounds cool. 19:33:57 in some cases I like that our puppet commits run out after a delay... 19:33:57 cool.. 19:34:08 in other cases, it's something urgent and you really would prefer fast. 19:34:37 idea -> it would be cool if zodbot published a link to ansible git diffs when we push so review can happen faster. 19:34:40 at least, be more public. 19:34:50 yep. I'm all for that. 19:34:58 nirik: aren't these the same most of the time (ie: urgent -> overlooked? :)) 19:35:05 sometimes. ;) 19:35:23 threebean: fedmsg? :) 19:35:43 pingou: https://gist.github.com/ralphbean/5963084 19:35:49 well if zodbot listened to fedmesg and published them 19:36:11 threebean: <3 #9 19:36:18 fedmsg would be ok with me too I guess... although not sure how interesting those messages would be to many fedmsg consumers. 19:36:36 but I guess they can filter them if they don't care. 19:36:50 lets see... other sysadmin stuff... 19:37:01 nirik: you don't want your desktop notifications of commit on ansible ? :) 19:37:06 smooge: any news on new bladecenter, bvirthost replacements or memory ordering? ;) 19:37:16 nope 19:37:25 pingou: might be amusing. I mostly just watch #fedora-fedmsg tho 19:37:28 I am getting a quote on memory. and I will be putting it in tomorrow 19:37:39 I have been spending most of my week on a GNOME hardware issue 19:37:39 smooge: the bvirthosts should be ready to order right? 19:37:42 yeah. ;( 19:38:02 nirik, I believe so I will do so after I get my allergy shot after this meeting 19:38:15 smooge: oh, also, where were we with that other ppc box? 19:38:25 completely unknown. 19:38:48 I will ping shaun on it 19:38:49 it was racked? or we don't even know that? 19:39:21 I thought it was racked but I can't find the email on it 19:39:40 what I thought was it was the other ppc that we are removing 19:39:44 oh yeah, and on the other ppc box... we were waiting for nfs network to get them added to koji.. 19:39:58 should finish that up if we can. 19:40:41 ok, any other sysadmin type stuff? 19:40:48 oh email ticket was put into system about bladecenter. 19:40:58 nirik, I need you to go look at fed-cloud03 19:41:12 smooge: ok, can do after meeting? 19:41:32 I believe its eth0 is now pointing to the internet but I can't get into it because well its uhm pointing to the internet with an 172 ip 19:41:42 nirik: we could always not use it for createrepo 19:42:16 dgilmore, ? with all the net noise not sure what you are talking about 19:42:20 dgilmore: yeah, although they have a lot of memory and will mostly be sitting there doing nothing, so it seems like createrepo would be something productive for them. ;) 19:42:32 smooge: the ppc01/ppc02 instances on the first new ppc box. 19:42:57 dgilmore, I believe it should be a quick edit and push to the puppet repo 19:43:11 the internal puppet repo for netapps not ours 19:43:22 smooge: they are on the nfs network on eth1 now? 19:43:31 I believe so 19:43:56 I tested pinging and such 19:44:13 cool, we can finish that up then. 19:44:25 #topic Upcoming Tasks/Items 19:44:25 https://apps.fedoraproject.org/calendar/list/infrastructure/ 19:44:31 by the way. I found our backup AIX knowledge person who can help us 19:44:36 anything upcoming folks would like to schedule or note? 19:44:39 smooge: excellent. 19:44:42 rbergeron, 19:44:45 sad that we need one. 19:44:51 ha! nice! 19:45:51 #topic Open Floor 19:45:59 any items for open floor? 19:45:59 ok I think we are going to need to set aside some time at Flock for a bunch of things 19:46:06 smooge: NO 19:46:17 rbergeron: NO? 19:46:19 smooge: yeah, we always have quite a list. ;) 19:46:26 sorry rbergeron you laid out your darkest secret about AIX and I jumped on it 19:46:36 because I am a rat bastard 19:47:04 smooge: i have a 13 year old ppc aix box if you need me to send it to you 19:47:25 see rbergeron you can use it to test the commands for our LPARs 19:47:33 and you are in PHX 19:47:43 its all win 19:47:46 i love when a plan comes together 19:47:54 heh 19:48:21 rbergeron: we love you 19:48:48 ok, anything else from anyone, or shall we call it a meeting? 19:48:53 ok for the Flock stuff I think it is mainly going to be not cool stuff :/. 19:49:12 * mdomsch will have to miss flock :-( 19:49:18 smooge: i have a list for you to do ;) 19:49:22 mdomsch: :( 19:49:25 :( 19:49:41 mdomsch: :( Thats very sad news. 19:49:47 mdomsch, I notice you mention it after I said it was not cool stuff. 19:50:00 smooge: not entire coincidental :-) 19:50:04 mdomsch: arf :( 19:50:04 that was a lie. It is really we are going to do it on the Aircraft Carrier Yorktown 19:50:16 woo! 19:50:19 and there is cake 19:50:25 punch and pie. 19:50:30 smooge: you know what they say about cake 19:50:32 and spoon 19:51:08 ok well I don;'t have anything 19:51:19 ok, will close out in a minute if nothing more. 19:52:29 thanks for coming everyone. 19:52:32 #endmeeting