18:00:09 #startmeeting Infrastructure (2015-07-09) 18:00:09 Meeting started Thu Jul 9 18:00:09 2015 UTC. The chair is nirik. Information about MeetBot at http://wiki.debian.org/MeetBot. 18:00:09 Useful Commands: #action #agreed #halp #info #idea #link #topic. 18:00:09 #meetingname infrastructure 18:00:09 The meeting name has been set to 'infrastructure' 18:00:09 #topic aloha 18:00:09 #chair smooge relrod nirik abadger1999 lmacken dgilmore mdomsch threebean pingou puiterwijk pbrobinson 18:00:09 Current chairs: abadger1999 dgilmore lmacken mdomsch nirik pbrobinson pingou puiterwijk relrod smooge threebean 18:00:10 #topic New folks introductions / Apprentice feedback 18:00:29 Any new folks like to introduce themselves ? 18:00:34 or apprentices with questions ? 18:00:57 Hi 18:01:03 I am aadityanair 18:01:10 from India 18:01:39 I was wondering how to get SSH access onto the machines 18:01:39 welcome, AadityaNair! 18:01:42 welcome AadityaNair. 18:01:59 are you more interested in sysadmin work or application development, or both? 18:02:27 Question: do we also provide support for apps inside infra? 18:02:27 sysadmin work is my prefernece 18:02:35 * threebean is here 18:02:40 * relrod is here 18:02:49 * rahulrrixe_ is here 18:02:52 * lmacken 18:03:18 * puiterwijk is here but checking something 18:03:22 jcvicelli: yes? you have an example? 18:03:26 Hi AdityaNair :) 18:03:30 * pcreech|work is here, but at nginx training 18:03:35 * prth is here 18:03:47 AadityaNair: great. see me in #fedora-admin after the meeting and I can get you added to our apprentice group to look around. 18:03:51 * ardian is here 18:04:41 nirik: ok, thanks. I will just sit back during the meeting. 18:04:52 Like koji for example 18:04:54 any other new folks with introductions or apprentices with questions? 18:05:00 jcvicelli: yes. We support koji. 18:05:31 And also develop? 18:05:31 nirik, yes 18:05:37 jcvicelli: mostly, yes :) 18:05:46 Cool, thx 18:05:55 jcvicelli: we are actively involved yeah... there is an upstream project tho seperate from us 18:05:59 jcvicelli: we play less of a role in koji development than in other things. it varies from service to service how much we're involved. 18:06:03 The list for easyfix https://fedorahosted.org/fedora-infrastructure/report/14 has some old tickets are they still relevant 18:06:24 ardian: likely yes. :) 18:06:49 I can try and go through them, but if they were not closed, they should be handled. 18:07:00 often someone starts working on one and wanders off. 18:07:06 I see 18:07:29 Looks like they are done but it's not clear what is missing 18:07:55 feel free to ask questions in the ticket. I get copies of every email to every ticket and can try and update status. 18:08:08 one more thing if the ticket says that it's owned by sysadmin-members 18:08:20 does that mean that you have to be in that group ? 18:08:31 no, thats just the default generic group that gets email about it. 18:08:38 some types of tickets go to specific groups. 18:08:44 thats just the generic one 18:09:16 ah ok, I am new and looking to find my way through the tickets 18:09:16 thanks 18:09:42 cool. I keep meaning to try and file some more easyfixes, but it's hard to come up with things that are really easy 18:09:59 but do ask questions and we can try and get you working on something you like. ;) 18:10:07 #topic GSoC student update - kushal 18:10:23 I don't see kushal around, but any GSoC folks want to update us on their projects? 18:11:17 On glittergallery, I am working on implementation of git ssh protocol. It is crucial for sparkleshare 18:11:25 Hi, I have figured out how to integrate the styles I have coded into askbot with the help of TheMonster which has given me trouble during the past weeks. 18:12:10 This week I have worked on mainly adding review process in fresque. I have made a simple commenting system and would be integerating with app. Also I have fixed some bugs on the git integeration. 18:12:27 So, during the next few days I will be working on integrating the styles into the test instance set up in openshift. 18:12:32 I implemented delete overlay. Last week after bug reports from pingou, I've been testing the app with high resolution wallpapers & fixed the bugs. I'm working on cropping wallpapers server side now as browser crashes for wallpapers > 3000*3000 with segments > 1000*1000 18:13:14 cool. Sounds like great progress all around. 18:14:04 #topic announcements and information 18:14:04 #info July apprentice email is out. Please reply by next monday - kevin 18:14:04 #info Some cloud instances moved from old -> new cloud - kevin 18:14:04 #info Fedora22 Jenkins builder setup in new cloud - kevin 18:14:04 #info people03 migrated to people01 - kevin, patrick 18:14:05 #info Ipsilon to prod scheduled for upcoming wednesday. Please test - Patrick 18:14:06 #info Making progress with the bodhi2 staging deployment https://admin.stg.fedoraproject.org/updates/ - lmacken 18:14:09 #info fun but slow progress on fedora-hubs design and implementation https://pagure.io/fedora-hubs/raw/develop/f/docs/diagram.png - mrichard / ralph / nyazdani 18:14:15 lots of info / status as usual. ;) 18:14:35 I had one discussion item: 18:14:35 #topic Outage announcements - kevin 18:14:37 lmacken++ on bodhi2. rad to see it up in staging. 18:14:49 I do want to point out the Ipsilon one if possible. Please test thoroughly this week :) 18:14:57 puiterwijk: can do 18:14:59 threebean: been working behind the scenes on the client/cli stuff yesterday... getting close ;) 18:15:05 nirik: real quick, how did you turn that ascii chart into a png? 18:15:06 lmacken++ indeed. bodhi2 goodness 18:15:08 nirik: Karma for lmacken changed to 7: https://badges.fedoraproject.org/tags/cookie/any 18:15:08 super 18:15:27 nyazdani: which chart? 18:15:38 nyazdani: ah, this tool -> https://apps.fedoraproject.org/packages/ditaa 18:15:54 nirik: sweet, thank you 18:16:19 this is in reference to this two files -> https://pagure.io/fedora-hubs/blob/develop/f/docs 18:16:25 s/this/these/ 18:17:02 cool. Looks nice. ;) 18:17:45 ok, on outage tickets. Right now we have a template in trac with a somewhat incomplete list of our services in it, and we fill it out for an outage with 'affected services' and 'unaffected services' lists. 18:17:57 we can somewhat get a more up to date list from apps.fedoraproject.org 18:18:13 but personally I wonder how much people care about 'unaffected services' 18:18:41 anyone have thoughts on how we best should handle them? 18:19:16 hm. not sure on the second part yet.. 18:19:55 but pulling the list of services dynamically from apps.fp.o/js/data.js could be cool. 18:20:28 hm. the list of unaffected services goes on-and-on at the end of the template. perhaps if it was just one comma-separated line.. 18:20:32 thing is, it might also not be complete... or some of them might not have urls. 18:20:39 then people who care about one service or another could search for the text of it and see that it's there.. 18:21:23 yeah, but if we only include affected services they could look and not see it there and know it's ok? 18:21:50 I guess the danger is that we might not think of everything affected and forgot to list something. 18:22:40 I guess I will look at how other places do things and see if we can come up with something clever. 18:22:50 #action nirik to check other projects for ideas 18:23:03 any other discussion items today? 18:23:33 #topic Learn about: fedorapeople.org - kevin 18:23:48 ok, I thought I would do a quick learn about on fedorapeople.org since we have been working on it. 18:24:38 fedorapeople.org is a server we have setup to allow any contributors to login and upload/distribute content from. 18:24:53 We just migrated it the other day from a rhel6/puppet managed host to a new rhel7/ansible managed one. 18:25:29 https://fedoraproject.org/wiki/Fedorapeople_Repos https://fedoraproject.org/wiki/Infrastructure/fedorapeople.org 18:25:57 contributors can login with their ssh key. 18:26:08 They can use a public_html directory to serve files out 18:26:29 Is it all done or you have still work to do? 18:26:30 They can use repos.fedorapeople.org / /srv/repos// to serve yum/dnf repos 18:26:47 They can use public_git to serve out git repos 18:27:00 jcvicelli: it's pretty much all done aside from fixing any issues people report. ;) 18:27:07 *is late 18:27:37 Also, fedorapeople.org gathers rss feeds from contributors and serves them out as planet.fedoraproject.org (soon to change to fedoraplanet.org) 18:27:57 * kinokoio_ is on a windows computer slow as hell 18:28:11 welcome kinokoio_ 18:28:42 We have quite a lot of data on fedorapeople, as it's one of the earliest services we offered. 18:29:08 There's about 460GB of content (most of it rpms) on there. 18:29:34 The server it's on is hosted at ibiblio. 18:29:48 Thats the high points I think. Any questions on the setup? 18:31:02 Alrightly then. ;) 18:31:05 #topic Open Floor 18:31:15 any questions, comments, announcements, puzzles? 18:31:25 Hi. I have something I would like to ask/request 18:31:33 sonalkr132: fire away 18:31:54 Until now glittergallery was using openshift for hosting 18:32:28 but now I need to implement ssh access for git. I will need some sudo access to limit ssh access 18:32:55 I also need access to .ssh/authorized_keys file. Which openshift doesn't provide 18:33:08 Which would mean that we move to VPS 18:33:25 per user keys? or ? 18:34:07 nirik: Point is I need a VPS, which we don't have. 18:34:26 well, we can set you up with a development instance in our cloud pretty easily. 18:34:50 that would be great. Whom should I talk to about this? 18:35:10 please file a ticket on it? there's a cloud instance request type 18:35:23 and say what os/size you need... 18:35:42 ohkay! Would that be all? 18:36:19 yep. we can get it setup... 18:36:31 thank you. I more thing 18:36:51 In our team we don 18:37:27 we don't have any one experienced with git server and stuff, so I was kinda having hard time. 18:38:08 do feel free to ask in #fedora-apps / #fedora-admin and we can help if we are able... 18:38:13 Our project is very similar to that of gitlab. So, does anyone know who can talk me up to a gitlab developer? 18:38:42 * nirik doesn't know any of them off hand. ;( 18:39:41 sonalkr132: pingou and bochecha are probably your best bet for advice on that, but they're likely off for the night now due to timezones. you may catch them tomorrow? 18:39:42 Thank you. I guess I can handle it. 18:40:18 nirik, was ticket https://fedorahosted.org/fedora-infrastructure/ticket/4779 already discussed? 18:40:47 threebean: thank you. I will talk to them 18:40:56 lmacken: so, f23 branching is next tuesday. I guess we won't have bodhi2 in production before then. :) Can we shoot for before alpha freeze? (28th?) 18:41:08 kinokoio_: yeah. we did. 18:41:38 kinokoio_: going to try and gather more data on with other projects do/best practices. If you would like to look for that info and add it to the ticket that would be great. 18:42:01 nirik: yeah, the alpha freeze is probably more realistic, since I'm moving apartments next week 18:42:28 lmacken: yeah. Please let me know if I can do anything to help it along (make instances, help out with ansible side, etc). 18:42:44 nirik: cool, will do. 18:42:45 I really want to get it in. :) Now also because I want to kill releng04/relepel01 from puppet. 18:42:56 nirik: I'll look for the info, maybe other distros or sysadmin channels would be a good place too ask 18:43:02 * lmacken nods 18:43:14 kinokoio_: yeah, I'd check debian/suse/other mailing list archives, etc. 18:43:26 we are getting down to the wire on puppet migrations. ;) 18:43:39 I hope to have them all done by flock one way or another. 18:43:54 nirik: need any help with the migrations? 18:44:51 kinokoio_: sure, but we are down to some very hard machines. ;) There's one ticket out there for migrating darkserver. 18:45:28 https://fedoraproject.org/wiki/Infrastructure/PuppetToAnsibleMigration and we should get 3 more in the next few days 18:45:54 ok, anything else for open floor? 18:45:58 or shall we call it a meeting? 18:46:07 i'll get into both when i'm on my pc 18:46:08 one thing 18:46:26 I should probably mention that I'm going to be afk july 13th-24th. 18:46:46 I wanted to mention that I was talking with pingou about moving the crawlers closer to the mirrors 18:47:08 threebean: have a great time away. :) Hope you are going somewhere fun... 18:47:09 so that if we have machines somewhere that we use them to crawl the local mirrors 18:47:15 adrianr_: I am not sure thats practical 18:47:52 many of the sites we have proxies don't have lots of resources 18:48:00 I doubt any of them have 48GB or whatever. 18:48:21 nirik: ok, good to know, was just thinking about what can be improved 18:48:42 so, you were looking for non NA/usa? 18:49:27 I know that the proxies are in different locations, but not exactly where, 18:49:48 to reduce long crawl times it might help to crawl closer to the mirrors 18:49:50 yeah, I don't think we really have any good remote sites for a crawler. Unless it could use a lot less memory 18:50:40 ok 18:50:50 there's one host in germany with some memory, but it's low on cpus. 18:51:16 good idea tho. ;) 18:51:22 we could try to crawl european mirrors from there, not so much in parallel, more serial 18:51:51 maybe instead of 30 at a time, only 8 and then we might can make this work with 8GB 18:52:16 we could give it up to 16GB, but not sure how many cpus. All are currently commited there. 18:52:24 we can discuss out of meeting I guess. ;) 18:52:34 sure, just wanted to mention it here 18:52:40 * nirik nods. 18:52:56 ok, if nothing else will close out in a minute. . 18:53:30 thanks for coming everyone! 18:53:32 #endmeeting