18:02:43 #startmeeting Infrastructure (2014-09-25) 18:02:43 Meeting started Thu Sep 25 18:02:43 2014 UTC. The chair is nirik. Information about MeetBot at http://wiki.debian.org/MeetBot. 18:02:43 Useful Commands: #action #agreed #halp #info #idea #link #topic. 18:02:43 #meetingname infrastructure 18:02:43 #topic aloha 18:02:43 #chair smooge relrod nirik abadger1999 lmacken dgilmore mdomsch threebean pingou puiterwijk 18:02:43 The meeting name has been set to 'infrastructure' 18:02:43 Current chairs: abadger1999 dgilmore lmacken mdomsch nirik pingou puiterwijk relrod smooge threebean 18:02:51 * lanica is here for the infra meeting. 18:02:55 oy 18:02:55 * tflink is here 18:03:00 hi 18:03:04 * puiterwijk here 18:03:06 neldogz is here 18:03:16 o/ 18:03:25 * lmacken 18:03:30 * pingou here 18:03:32 * bitlord here, listening \o ;-) 18:04:17 * danielbruno here 18:04:24 #topic New folks introductions and Apprentice tasks 18:04:29 any new folks like to introduce themselves? 18:04:34 or apprentices with questions or comments? 18:04:50 * oddshocks arrives enveloped in a cloud of crows 18:04:57 * relrod arrives too 18:05:03 ! 18:05:46 >:} 18:06:09 danielbruno: feel free to just chime in. we don't use that meeting protocol thing. 18:06:34 no agenda ? 18:06:38 nirik, I saw yout reply on the ticket about the planet to track broken feeds 18:06:55 ssf87: yes there is. It was sent to the mailing list... 18:07:02 I would ike to know if I need some permission 18:07:17 to access .planet files 18:07:27 danielbruno: you shouldn't. :) everyones /home/fedora/*/.planet files should be readable. 18:07:41 great! 18:07:44 the planet job runs as nobody so if it can read it you should be able to too 18:08:10 nirik, thank you :) 18:08:22 no problem, do let me know if you run into any problems with it. 18:09:18 any other new folks or apprentice questions/comments? 18:09:52 #topic Applications status / discussion 18:10:06 any application news this week? threebean / lmacken / pingou / oddshocks ? 18:10:15 we're having issues with fedmsg load on datanommer right now 18:10:16 http://threebean.org/fedmsg-health.html 18:10:31 I did a couple of bodhi updates in production this week. Nothing too exciting, mainly EL7 related stuff. 18:10:35 threebean: is that db on rhel7 yet? or still 6? 18:10:37 I'd like to at least throw in the public again that fedoauth will be deprecated in the future. not yet sure when 18:10:38 it's a result I think of us outgrowing our current one-big-postgres-table setup and we're researching alternatives 18:10:46 I've been working on a couple of requests from gnokii for nuancier, review pending 18:10:59 I also have some reviews pending for pkgdb2, closing some tickets 18:11:12 #info fedmsg load issues with datanommer, being worked on. 18:11:25 #info some bodhi1 updates in production around epel7 stuff 18:11:28 and I added support to send the reminder emails to multiple addresses on fedocal, merged and planned in the next release 18:11:31 nirik: still on 6.5 18:11:33 threebean: it'd be real interesting to see those benchmarks that you wrote earlier between RHEL 6 & 7 18:11:43 #info fedoauth will be depreciated at some time to be scheduled yet. 18:11:45 #link https://lists.fedoraproject.org/pipermail/infrastructure/2014-September/014863.html 18:12:02 #info some nuancier patches pending review. 18:12:23 threebean: wonder if this would be a good time to move it to 7? or it needs more than just a newer postgres? 18:12:28 * threebean nods 18:12:36 let's move it up this afternoon. 18:12:47 I can update moksha and the collectd stuff on busgateway01 at the same time. 18:13:15 If anyone uses HRF (hrf.cloud.fp.o) for anything, please switch to datagrepper instead. I want to deprecate HRF because datagrepper can do everything HRF was made to do now. 18:13:37 so, would we want to spin a new 7 one up and transfer, or just save the db off and destroy the 6 one and recreate it as 7? 18:13:50 the first option sounds safest :) 18:14:08 #info hrf.cloud.fedoraproject.org depreciated. Please don't use it anymore, use datagrepper. 18:14:24 threebean: yeah, just needs allocating ip's, etc... more infra work, but not too big a deal. 18:14:48 taskotron production is making progress - initial systems are set up, proxies are configured, seems to be mostly working 18:15:04 still need to get monitoring and backups set up 18:15:20 tflink: saw the monitoring ticket, but haven't had a chance to do anything with it. 18:15:28 would like to see it funcitoning for several days and a couple of fixes in place before turning off autoqa 18:15:28 might be a good place for a new person to jump in. ;) 18:15:33 yeah. 18:15:48 SOPs have been written for taskotron and resultsdb 18:15:57 #info taskotron almost in production. Needs monitoring and backups and a few days of good smooth operation. 18:17:20 cool. Any other applications news? 18:17:30 anything we want to make sure to try and get done before beta freeze? 18:18:06 MM2? :D 18:18:08 * pingou ducks 18:18:11 F21 RC1 AMIs are listed on my fpeople page, still debugging 32 bit base and 64 bit atomic: https://oddshocks.fedorapeople.org/ 18:18:18 but 64 bit base are there 18:18:33 oddshocks: nice. Should those get updated in the website? 18:18:42 pingou: ha. 18:19:11 pingou: I hear you're volunteering to get it done by then? :) 18:19:11 nirik: I think robyduck is waiting to update the links until i have the other two sets working, but i'm not sure if that's the same thing 18:19:22 we might manage to have anitya running before beta 18:19:24 oddshocks: ok, sounds reasonable. 18:19:31 but that's not quite part of our infra anyway :) 18:19:42 puiterwijk: I was merly proposing you :-p 18:19:49 also, as a side note, I put f21alpha cloud image into our cloud... so if we need any f21alpha instances it should be ready. 18:20:02 we could add a jenkins one perhaps. 18:20:46 which reminds me... pingou: what was the conclusion about making jenkins a more supported/supportable service? no go since it's not packaged for rhel? or ? 18:21:11 nirik: not packaged for rhel was/is the biggest blocker 18:21:32 yeah. wonder if even 7 wouldn't work... it should have new enough stuff I would think 18:21:54 depends on jenkins' deps, which I suspect is big 18:22:02 yeah, jenkins dep tree is quite big 18:22:04 the docs folks were looking at a jenkins plugin for publishing and serving docs. 18:22:18 maybe we should get someone to just go ahead and package it all 18:22:28 yes, I read the backlog of this 18:22:41 puiterwijk: that would imply we are comitted to maintain it 18:22:50 pingou: yeah 18:23:10 pingou: do you recall who we were talking to about it at flock? 18:23:29 nirik: I remember the face, but not the name 18:23:32 :/ 18:23:34 yeah, same here. ;( 18:23:57 oh well, we don't need to do anything right now, but something to think on. 18:24:13 #topic Sysadmin status / discussion 18:24:24 hi 18:24:26 #info CVE-2014-6271/CVE-2014-7169 (Bash issues): Patches for 6271 and workaround for 7169 applied on high profile servers 18:24:28 so, we left freeze yesterday... just in time for some security update fun. ;) 18:24:34 yay! 18:24:40 hey smooge 18:24:49 puiterwijk: you also updated mediawiki right? 18:25:00 nirik: yup. we're now at the latest one, released yesterday 18:25:13 #info Mediawiki upgraded to 1.19.19 18:25:38 yay! 18:26:15 \0/ 18:26:15 cool. 18:26:28 and I'm packaging the 1.23 series 18:26:41 (since we need to upgrade before May 2015, since that's EOL for 1.19) 18:27:16 yep. 18:27:32 I've been working thru backlog of tickets that landed when I was traveling... 18:27:43 hopefully we will be back to normal on those before too long. 18:27:53 I'd also like to look at scheduling a mass reboot cycle next week. 18:27:58 Probibly wed or so... 18:29:00 #info mass reboot likely next week to catch up on updates 18:29:17 #topic nagios/alerts recap 18:29:26 * nirik looks for the url again. 18:30:04 .tiny https://admin.fedoraproject.org/nagios/cgi-bin//summary.cgi?report=1&displaytype=3&timeperiod=last7days&smon=9&sday=1&syear=2014&shour=0&smin=0&ssec=0&emon=9&eday=4&eyear=2014&ehour=24&emin=0&esec=0&hostgroup=all&servicegroup=all&host=all&alerttypes=3&statetypes=3&hoststates=7&servicestates=120&limit=25 18:30:04 puiterwijk: http://tinyurl.com/l3vjae8 18:30:11 nirik: ^ 18:30:17 and again puiterwijk comes through. 18:30:21 beat me to it. ;) 18:30:57 nirik: or better: http://da.gd/fednagios 18:30:59 so, the datagrepper issues we are aware of 18:31:45 not sure about the collab mail queue. I don't recall seeing those? 18:31:50 so must have been warnings 18:31:57 * oddshocks read that as http://da.gd/fadingos 18:31:57 packages03 might need some more memory :\ I spent a little bit of time this week poking some unicode issues with the xapian db, but haven't looked at prod 18:32:25 lmacken: yeah, it's gotten stuck a lot lately. It also had a issue this week where OOM killed glusterd. ;( 18:32:31 we can bump it up some more. 18:33:01 8gb -> 12? 18:33:14 that sounds fine 18:33:47 also, packages.stg can't get to dl.fp.o:80 for some reason. Need to add a firewall rule? 18:33:53 * nirik can do so after the meeting. 18:33:54 it can ping it 18:34:12 is it using internal or external ip? 18:34:36 I think it was hardcoded to download03.phx2 18:34:37 lmacken: yeah, stg -> prod is denied with internal IP now, thanks to threebean's blanket rule 18:34:56 okay, cool 18:34:57 so it'd need to use external IP to have any chance, or a seperate firewall rule 18:35:07 yeah. 18:35:19 (I vote for external IP, I rather like the stg->prod firewall rule) 18:35:29 if external works, sure. 18:37:01 also, re: nagios... I am going to kill unbound-telia01 soon... and possibly kill mirrorlist-serverbeach too. (Although I might make it a rhel7 and see if it's any happier that way) 18:37:26 nirik: update: apparently those 64 bit base amis _are_ on the website, and roby will add the rest once i figure out what's wrong with them 18:37:34 oddshocks: great. :) 18:37:47 any other nagios related or sysadmin related stuff before we move on? 18:37:56 HVM/atomic stuff is proving to be a bit of a learning curve for me, still working out what options need to happen for the image to boot 18:38:28 having to add a bit of fedimg code to accomplish that special stuff, and since certain instance types can and can't be HVM, and some instance types can and can't be 32 bit, it's a bit of a challenge but i'm working through it 18:38:40 * oddshocks done derailing the sysadmin section 18:38:47 yeah, I think the atomic stuff is a learning curve for everyone. :) 18:39:24 I wonder if I could set up a download01.stg.phx2.fedoraproject.org? 18:39:36 we could if needed sure. 18:39:49 seems kinda a waste unless it's different any 18:40:01 well I figured it doesn't need to be hardware 18:40:26 just an virt for stg stuff to have 1:1 parity to 18:40:31 sure, but it would just mount the same stuff? it just seems like it wouldn't be too usefull... and would take up memory/cpu/etc 18:41:25 but I guess if we can't get things talking to the prod ones we could. 18:41:49 #topic Upcoming Tasks/Items 18:41:50 https://apps.fedoraproject.org/calendar/list/infrastructure/ 18:42:01 anything upcoming anyone would like to note or schedule? 18:42:28 hopefully a real fix for CVE-2014-7169, while I'll deploy once it arrives 18:42:36 yeah. 18:42:36 while=which 18:43:22 * nirik nods 18:43:29 #topic Open Flood 18:43:38 * nirik typos, runs with it. 18:43:39 nirik: new name for the open floor? :) 18:43:41 anything to flood? ;) 18:43:44 * puiterwijk likes it 18:43:46 * pingou heads for dinner, ttyl :) 18:43:58 pingou: enjoy 18:43:58 will do :) 18:44:12 (really this is open floor, so bring up any other topics anyone would like to discuss) 18:44:15 lots of work to do until beta freeze... ;) 18:44:24 yeah... 18:44:35 I'd really like to get more rhel7 and ansible migrations done 18:44:39 random solicitation: i've been working on an update to the landing page at https://apps.fedoraproject.org 18:44:57 if anyone wants to help fill in the last bits of content, that would be a help -> https://github.com/fedora-infra/apps.fp.o/ 18:45:53 cool. 18:47:45 ok, if nothing else will close out in a minute. 18:48:19 Thanks for coming everyone! lets continue in #fedora-admin, #fedora-apps and #fedora-noc. 18:48:21 #endmeeting