18:59:58 #startmeeting Infrastructure (2012-01-26) 18:59:58 Meeting started Thu Jan 26 18:59:58 2012 UTC. The chair is nirik. Information about MeetBot at http://wiki.debian.org/MeetBot. 18:59:58 Useful Commands: #action #agreed #halp #info #idea #link #topic. 18:59:58 #meetingname infrastructure 18:59:58 #topic Robot Roll Call 18:59:58 #chair smooge skvidal Codeblock ricky nirik abadger1999 lmacken dgilmore mdomsch 18:59:58 The meeting name has been set to 'infrastructure' 18:59:58 Current chairs: Codeblock abadger1999 dgilmore lmacken mdomsch nirik ricky skvidal smooge 19:00:40 * CodeBlock here 19:01:02 * LoKoMurdoK here 19:01:05 * nirik waits for folks to trickle in 19:01:20 hello 19:01:45 * lmacken 19:01:46 * wsterling here 19:02:15 hi 19:03:01 hi 19:03:22 * jac1bat0 is here 19:03:46 * abadger1999 here 19:03:52 #topic New folks introductions and Apprentice tasks. 19:03:58 * pingou here 19:04:12 any new folks want to say hi and introduce themselves or apprentices want to talk about tickets? 19:04:21 Hello - I'm Dave Parfitt from Buffalo, NY. I write software and fix things. I have been using Linux on and off since the 90's. 19:04:37 welcome metadave 19:04:43 thanks 19:05:40 any other new folks? 19:06:50 ok, moving along... 19:06:54 #topic 2 factor auth status 19:07:00 hey metadave! 19:07:11 hello abadger1999 19:07:15 any news here? /me looks for herlo 19:08:23 * nirik moves on 19:08:29 #topic reworking staging 19:08:44 We have started in on this... averi has done some work here. 19:09:07 we will be working on it more next week/coming weeks 19:09:10 sorry 19:09:15 no worries. 19:09:22 was in another window in my brain 19:09:56 hopefully sometime in the next few weeks we can nuke the stg branch. 19:10:04 I also intend to re-install a few stg machines. 19:10:27 nirik: and if you want to give averi access - he seems to want to wotrk on more of them 19:10:39 ok, we can take a look at that... 19:11:12 #topic Mass reboot outage later tonight. 19:11:20 yay! 19:11:21 fun! 19:11:26 https://fedorahosted.org/fedora-infrastructure/ticket/3113 19:12:22 we have our class A and B reboots tonight. 19:12:26 so there will be an outage. 19:12:48 http://infrastructure.fedoraproject.org/infra/docs/massupgrade.txt is our SOP here. 19:14:27 Action for that will likely be in #fedora-noc 19:14:54 #topic Applications status / discussion 19:15:09 So, this worked somewhat last week, so lets try again. ;) 19:15:35 Any application owners have items to discussion about their apps? 19:16:38 we have a fas release scheduled for 2012-02-07 19:16:55 the audio captcha can already be tested on stg 19:16:58 * abadger1999 checks what his email said 19:17:09 pingou: great! 19:17:19 thanks to abadger1999 ;-) 19:18:46 nirik: oops, I announced Feb 2 as cutting the final release. 19:18:59 ok. 19:19:05 * nirik fixes 19:19:13 I think I slipped development stuff a week, but only slipped release a few days 19:19:35 fair enough 19:20:22 lmacken: any news on bodhi and community? 19:20:34 sounds like we had a discussion on list about urls, but did we reach a conclusion. 19:20:42 nirik: I'm working on polishing up a new bodhi release containing the FESCo proventester policy change 19:20:59 it was a 2 line change, but the test suite needs some love to get it back up and running 19:21:04 k 19:21:08 as far as community/packages, we need to get it deployed to production 19:21:19 yeah 19:21:21 * lmacken has been doing very careful bugfix updates 19:21:37 and that's all I got 19:22:05 for deployment, I was thinking 2 servers running the app and 1 db server. 19:22:16 then we need to figure out urls. 19:22:20 sounds good. I completely seperate db server? 19:22:24 s/I/A/ 19:22:34 right now the only thing that uses a db on community01.dev is tagger 19:22:48 do we have a timeline for that deployment? 19:22:48 yeah, so that way this is more self contained... but I guess thats open to discussion. 19:22:56 sorry if I missed it - just trying to catch up a bit 19:23:12 do you see any issues with 2 app servers? 19:23:17 * lmacken doesn't 19:23:25 lmacken: Clue in me on xapian -- is it a db? Or backed by a db? Or files? 19:23:27 as soon as those exist, I can deploy it. 19:23:41 abadger1999: xapian is it's own file-based db 19:24:43 lmacken: and should we use our existing caching setup? ie, proxies -> haproxy -> varnish -> app1/app2 19:24:45 lmacken: do both app servers have a copy of it - or do they get synced out from a common source? 19:24:48 lmacken: and community communicates over the network to it? 19:24:52 nirik: that's fine with me 19:25:02 nirik: do we really want to put this all behind the proxies? 19:25:07 skvidal: good question. I haven't thought that part over. 19:25:09 might make it even speedier (is that a word) 19:25:28 abadger1999: nope, it's a local db currently 19:25:29 skvidal: well, it does help for geographic diversity... 19:25:48 nirik: except that we lose some of the proxies when things go sideways in phx2 19:25:51 lmacken: But does it have to be local files? 19:25:58 ok, so we have an indexer that runs every night... we can then have it sync the xapian db to both of our app servers 19:26:02 or is it tcp/ip capable? 19:26:15 abadger1999: I'm not positive, but I think it's file based only... 19:26:19 skvidal: sure, but would direct be any better? I guess wouldn't be subject to proxy failure. 19:26:23 nirik: I guess I was thinking - since this app doesn't tie to any other resource directly (no nfs, for example) maybe this is the time we model it for the future 19:26:41 skvidal: yeah, but then it's SPOF on phx2. 19:26:43 nirik: and build it up as an silo 19:26:58 nirik: why not stick it at ibiblio? 19:27:14 skvidal: it needs to index/hit koji/pkgdb/bodhi I think? 19:27:18 would be slow from remote... 19:27:34 nirik: oh - I thought the indexing happened locally and was pushed over to the apps 19:27:40 my misunderstanding, then. 19:27:43 how about this: 19:28:05 I'll start a thread on deployment on the list and we can hash out what questions we all might have and best place to put it, etc. 19:28:19 skvidal: the indexing is (for the search) but not the querying of package's info ( lmacken did I got it right ?) 19:28:44 nirik: sounds like a plan - would be handy to know the structure better 19:29:24 pingou: sounds correct :) 19:29:26 lmacken: hate to do this to you - but could you write up a small description of the server intedepdnencies? 19:29:33 skvidal: sure 19:29:40 lmacken: nothing complicated - just want to make sure I know where the bits matter 19:29:46 * nirik nods. that would be great. 19:29:50 skvidal: of just community01.dev stuff? 19:30:06 lmacken: of what we've discussed here, yeah - but ideally both packages and tagger 19:30:13 (which I assume falls under community01.dev) 19:30:22 k, will do 19:30:27 also, as a side idea: perhaps we get someone else to deploy it to new machines. ;) Just so we have someone able to do so in case lmacken isn't around and fix any problems with the sop? 19:30:39 that's totally fine with me ;) 19:30:48 nirik: +1 19:30:52 I'm on board with doing that 19:30:53 it /should/ be as simple as just running puppet 19:31:06 yeah. ;) 19:31:10 lmacken: if only puppet was as simple as running puppet :) 19:31:13 ok, any other applications news? 19:31:21 I did write a sop for it too http://infrastructure.fedoraproject.org/infra/docs/fedorapackages.txt [-] 19:31:27 do we have an actual updated list of our apps and who's point on them? 19:32:02 http://fedoraproject.org/wiki/Infrastructure/Services I guess. should get updated. 19:32:30 oh right, I forgot about that page.. I'll add packages & tagger 19:32:38 lmacken: thanks. 19:32:42 * nirik didn't know of it either. 19:32:47 * lmacken made it a long time ago 19:32:59 #topic Upcoming Tasks/Items 19:33:23 I have a list in my email agenda... is it worthwhile to spew it to the meeting too? 19:33:35 dgilmore will be in phx next week... 19:33:38 nirik: hey, just read backlog, I confirm what skvidal said, I can work on more hosts if you want me to. 19:33:59 averi: great, we can take a look after meeting. ;) 19:34:19 nirik: I am about to leave, will be back tomorrow after the outage! 19:34:31 will poke as soon as I login :) 19:34:31 If anyone can think of anything for dgilmore to do while out in phx2, we have a ticket: 19:34:33 https://fedorahosted.org/fedora-infrastructure/ticket/3084 19:34:37 nirik: yeah, need to finalise the list of things to do tehre 19:34:39 averi: no worries. Enjoy 19:34:55 thanks! later 19:34:58 did the ppc box turn up? 19:35:05 smooge: ^ 19:35:18 s/turn/burn/ 19:35:19 * skvidal hopes 19:35:42 dgilmore, box is there 19:35:52 skvidal: maybe it will burn up 19:35:56 smooge: gracious 19:36:03 also, we want to re-install sign-vault01... however I see it's warnetee ends 2012-04-11 19:36:17 nirik: hrrm 19:36:19 nirik: womp womp 19:36:29 I have now gotten all backups off backup01 19:36:35 nirik: replacement in 2 days? 19:36:35 I think we can get another box or an extension on that warranty 19:36:38 who's warnetee ends 2012-06-17 19:36:59 smooge: did we plan for a replacement box for that one? I can't recall 19:37:22 the reason i didn't get a warrantee extension was that it was a test box that f13 got that made production. 19:37:26 nirik: im going to do a equalogic firmware update while there also 19:37:33 dgilmore: and junk04? 19:37:41 dgilmore: cool. Does that require an outage? 19:37:42 I then put in a replacement for first q of next year.. but not sure if that got approved 19:37:54 nirik: yes, need to take the storage down to do it 19:38:16 skvidal: is that the one thats broken currently? 19:38:19 dgilmore: ok, we should schedule that outage then. 19:38:35 nirik: yeah, i was thinking next thursday night 19:38:41 proposal: repurpose backup01 as a new sign-vault02 19:38:41 should only take 15 mins 19:38:56 then we can extend it's warnetee hopefully. 19:39:07 and we have sign-vault01 around if something blows up 19:39:50 nirik: works for me 19:40:17 one question I had was: will backup03's tape drive read backup01 tapes? if so, we should save old tapes for a while in case we need them. 19:40:57 Anyone else have anything upcoming they would like to discuss? 19:40:59 nirik: backup03 has LTO4 drive? 19:41:47 LTO-5 19:42:10 LTO-5 (Half Height) it says 19:42:10 nirik: yeah it will be able tor ead LTO4 tapes 19:42:17 ok, great. 19:42:43 #topic Meeting tagged tickets: 19:42:44 https://fedorahosted.org/fedora-infrastructure/report/10 19:42:58 .ticket 3094 19:43:02 nirik: #3094 (Assigning a default acl to the pkgs repo) – Fedora Infrastructure - https://fedorahosted.org/fedora-infrastructure/ticket/3094 19:43:05 oh yeah, this got addressed. 19:43:12 * nirik will remove meeting keyword. 19:43:18 abadger1999: do you need any more info on that one? 19:43:40 nirik: Nope, We can even EasyFix this one -- it's all in one method. 19:43:46 I'll update it 19:44:28 cool. thanks. 19:44:35 #topic Useless Stats 19:44:45 I just had some useless stats I thought I would throw out: 19:45:06 we are down to just 16 RHEL5 instances now. Which is just over 10% of our total. 19:45:21 we are down now to 180 tickets. 19:45:45 #topic Open Floor 19:45:51 anyone have anything for open floor? 19:45:59 New infra-related channel #fedora-apps 19:46:14 ah yeah. 19:46:35 #info new #fedora-apps channel for application discussion, design, and issues. 19:46:52 We'll have some fedora-infr-created web app meetings and discussions in there. 19:47:06 \ó/ 19:47:32 You can still use #fedora-admin for discussion as well; at least until the traffic for those discussions grows :-) 19:48:14 * nirik sees #fedora-apps as more for design/early discussions or lengthy ones vs admin being for quick issues/bugs/fixes/notifications. 19:48:22 19:48:48 anyhow, anyone have anything else? or shall we call it a meeting? 19:49:00 call it a day 19:49:50 ok, thanks for coming everyone! 19:49:53 #endmeeting