16:03:44 #startmeeting 16:03:44 Meeting started Fri Oct 26 16:03:44 2012 UTC. The chair is johnmark. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:03:44 Useful Commands: #action #agreed #halp #info #idea #link #topic. 16:03:56 #chair Technicool 16:03:56 Current chairs: Technicool johnmark 16:05:12 pasted from email: 16:05:16 Today we will discuss 16:05:16 1. outcome of server migrations 16:05:16 2. recommendations for developer collaboration services - github, et al. 16:05:19 3. wiki madness - the wiki needs help. It's a bit of a sprawl 16:05:21 4. case study project 16:05:31 #topic server migrations 16:05:33 ooh@#4 16:05:36 I think we're mostly done here 16:05:41 :) 16:05:50 Technicool: that's where we ping semiosis. I think 16:06:13 i just liked the sound of it 16:06:19 heh :) 16:06:32 anyone notice server oddities? anything we should be doing differently 16:06:50 we will be closing out everything on iWeb shortly and moving the rest of the stuff over to rackspace 16:07:33 do we have any sort of strategery in place so that in the event of another failure, we can recover within hours or minutes? 16:08:09 as far as the site itself, haven't noticed anything in particular 16:08:36 STRATEGERY! who do you think this is? 16:08:47 Why my middle name is... 16:08:50 not strategery 16:09:10 that sounds like standing up a mirrored server on rackspace 16:09:12 mission accomplished... 16:09:21 and implementing some type of failover 16:09:28 ok, we need to make sure it is a gen1 server then 16:09:30 That's what automagic backups are for. 16:09:30 which I know nothing about 16:09:35 Technicool: yup, exactly 16:09:36 i can take care of the mirror and backups 16:09:41 you r0x0r 16:10:08 #action Eco to handle server mirroring and image backup scheduling once iWeb migration is complete 16:10:08 Technicool: is that all we need? or does rackspace also offer some type of hot HA? 16:10:49 they have a load balancing service, you need to pay for their "premium service" but the cost wasn't too high 16:10:50 or would we just need to use some heartbeat tool 16:10:54 ok 16:11:39 that also means making sure the data is consistent between the two 16:11:41 the load balancing includes failover, I assume heartbeat which means they probably don't have heartbeat for some reason just to make things interesting 16:11:55 hrm... pretty sure I've heard of a tool that does replication... let me look it up 16:12:07 rsync? 16:12:10 heh 16:12:11 LOL 16:12:12 :P 16:12:20 im here all week folks 16:12:56 ok, can you find out how much it costs? probably better to farm out as much of that as possible 16:13:08 so we're not always in the position of maintaining crap 16:13:17 That's what puppet's for. 16:13:26 * johnmark is looking to delegate as much as possible away from us 16:13:33 i can work up a total cost but it recall i never winced looking at the prices 16:13:35 JoeJulian: that, too 16:13:42 Technicool: a good sign :) 16:13:51 otherwise, I think we're set. 16:14:12 IIRC, it would be about 3X the price of the server instances we are using now, so still better than the competition 16:14:28 JoeJulian: if you want to check out puppet recipes, I won't turn you down :) 16:14:30 and the less anyone here needs to do to manage it, the better 16:14:36 cool 16:14:37 ok 16:14:50 next topic 16:14:54 unless there's something else to add here 16:15:00 #topic dev collab 16:15:10 there are all sorts of services we can use here 16:15:35 github, bitbucket, a few others 16:15:57 that allow us ot use vanity URLs and our own logos/branding 16:16:56 in the spirit of minimizing our work, we'll compare online services to DIY 16:17:30 but I gotta tell ya, online services we pay for sounds awfully attractive at the moment 16:17:37 which is not to say that's what we'll do 16:18:08 if we can get support for our stuff from engineers in-house, then the weighting is equal, I think 16:18:24 that should read "IT folks" not engineers 16:18:35 ah, ok 16:18:45 ie. not us 16:19:01 but this is something I'd like to have up on gluster.org before the end of the year 16:19:11 because I'm discovering new projects each and every day, and we don't have a place for them 16:19:23 im ok with the RHIT folks if we can get it done the way we need within a decent timeframe 16:19:50 pong 16:19:53 #action jmw to draw up comparisons between online services and self-hosted 16:19:57 semiosis: yeehaw 16:19:58 * semiosis catches up 16:20:13 semiosis: greetings, salutations, etc. etc. 16:20:15 otherwise, the pace of what happens here is fairly rapid, so that may mandate that we need to work up our own solution with an outside vendor 16:20:25 ok 16:20:37 Technicool: agreed 16:20:56 I'll send around whatever comparison points I can find 16:21:14 if any of you have a favorite online service for dev collaboration, please send it my way 16:21:53 ok, anything else before we move to the next topic? 16:22:41 #topic wiki mania / wiki madness 16:22:53 * Technicool quakes with fear 16:23:13 this is something we'll have to think about, but we're starting to get to the point where there are too many projects on the wiki competing for space on the front page 16:23:32 this is a natural by-product of doing more stuff 16:23:54 but we need to think about redesigning the front page so that people can find the information they want/need 16:24:07 Technicool: I'm inclined to ping garrett for his expertise 16:24:25 #action JMW to ping Garrett for design help 16:24:26 johnmark, yes, we should get him onboard asap if he is done with the ovirt project 16:24:31 word 16:24:36 please include me in any meetings with him 16:24:41 you bet 16:24:55 no one has ever accused me of being a designer 16:24:59 :) 16:25:23 ok 16:25:23 from an osas stance, it would also be good to brainstorm what types of effort are reusable across projects 16:25:33 Technicool: excellent suggestion 16:25:54 ok, last topic 16:26:04 #topic case studies 16:26:18 so we're getting a case study done with the Rock the Vote project, thanks to our fine furry friends at the OSL 16:26:19 * Technicool ear lifted, face aglow 16:26:29 and we need more, more, MOAR 16:26:30 very nice 16:26:55 I've never been much for studying. 16:26:57 at some point, we discussed throwing together a form that people could fill out and describe what they're doing with GlusterFS 16:27:01 JoeJulian: heh heh :) 16:27:09 JoeJulian, i think here we made someone else do the studying 16:27:14 semiosis: I think you talked about doing something along those lines 16:27:20 * johnmark looks at semiosis 16:27:32 Technicool: exactly! 16:28:08 anyhoo, if someone wants to put up a braindead form, such as on survey monkey or elsewhere, I would welcome it 16:28:31 ick @ survey monkey but intent noted 16:28:42 and I am, of course, available for help and consultation 16:28:53 johnmark, what is the purpose exactly of the form? 16:28:59 Technicool: could you recommend something else? i've heard of survey gizmo & polldaddy as well but never used any 16:29:03 "are you using gluster and how"? 16:29:13 #action jmw to find suck... er volunteer for case study project 16:29:28 Technicool: get people to fill out what they're using GlusterFS for, what their setup looks like, etc. et.c 16:29:39 semiosis, not offhand i just didn't like the trolly feel of surveymonkey 16:29:46 and then we'd follow up with further questions and churn out some type of thing on the web site 16:29:59 semiosis: I've heard of all sorts of things, I just can' tremember what they were 16:30:02 johnmark, gotcha....we discussed something similar awhile back 16:30:09 semiosis: some people use google docs for this. 16:30:18 semiosis: I'm not recommending it, just noting that ti's been done 16:30:22 interesting 16:30:32 it would be nice if we could have users fill this out themselves via the wiki 16:30:54 Technicool: good point. can you do forms with mediawiki? 16:30:58 then we can capture and poll the data directly without having overhead 16:31:05 Technicool: if not, we could just hand them a template and tell them to use it 16:31:11 you can, but we need to modify the install i think 16:31:12 and hope they don't screw up the template 16:31:12 :) 16:31:20 Technicool: cool 16:31:48 either way, i think this is a very powerful tool 16:32:03 semiosis: any ideas on what we should use? if we can get away with mediawiki, might as well use the tools we have 16:32:07 even if they're not perfect 16:32:09 to be able to show specific use cases, on specific hardware 16:32:16 yup 16:32:42 i think alot of people avoid Gluster just based on not knowing if it will work for them 16:32:54 and, i personally would like to see a list of fail cases 16:32:59 Me too 16:33:04 yup 16:33:28 hmm... "fail studies?" 16:33:39 I think that's supposed to be called bugzilla 16:33:42 would let us utilize a popular meme 16:33:43 then we can show over time that we either addressed the issue and turned it around to success, or else can say "well, you can try it if you want, but others already did and died according to our cool data" 16:33:46 heh 16:33:50 time is the most scarce resource... if we can get this going with less effort using a 3rd party service that gets my vote 16:33:58 JoeJulian: in a perfect world, yes 16:34:07 semiosis: cool 16:34:19 semiosis: I think that's a good point 16:34:33 JoeJulian, bugzilla gives you the parts, but case studies give you the STORY! 16:34:38 * Technicool waxes poetic 16:34:58 * Technicool likes the cut of semiosis jib 16:35:00 semiosis: but then it has to be a service that allows for easy migration to our web site 16:35:04 either via RSS or something 16:35:08 leave my jib outta this :{ 16:35:11 :P 16:35:13 jiblets 16:35:16 If you're throwing money at the problem... what about offering grants to students to survey and write case studies? 16:35:17 semiosis: ha! 16:35:32 JoeJulian, genius! 16:35:38 JoeJulian: +11111!!11ONEELEVEN 16:35:42 JoeJulian: interesting. there are things like mechanical turk from amazon for this 16:35:48 and other services 16:35:56 i also like the thought of using Mechanical Turk 16:36:06 ok. 16:36:18 if we can find a sort of repetitious job 16:36:26 so if someone can volunteer to investigate the online service for creating a form 16:36:27 i dont think mturk is good for this 16:36:30 mturk -1 16:36:37 offer 50 cents for each iteration, the turk monkeys would be all over that like white on rice 16:36:45 and someone else can look at mechanical turk alternatives 16:36:47 then we're set 16:36:49 semiosis, not for this, but if there is somethign else 16:36:49 semiosis: ok 16:36:59 Technicool: well yeah for lots of other things... 16:37:00 like re-importing data from said 3rd party service 16:37:14 waaaaay too complicated 16:37:25 semiosis, yes....i used mturk to refinance my house and it was a disaster 16:37:42 semiosis: I remember hearing about some service where you farmed out small jobs 16:37:45 but the name escapes me 16:37:57 Technicool: haha.... ouch! 16:37:59 semiosis, crackers use it to get past captcha so there has to be something fun we can do with it 16:38:09 :) 16:38:10 if we use a 3rd party survey service we'll likely get an XML doc or something we can use a script to get into mwiki easily 16:38:11 awesome 16:38:18 semiosis: that's what I assume/hope 16:38:22 mturk is good for spotting nipples in photos... 16:38:38 * Technicool points a browser toward amazon.... 16:38:39 semiosis: not that you have any previous experience with that 16:39:05 * Technicool digresses before he gets in real trouble 16:39:07 mturk, find me ALL the... 16:39:33 mturk, type in a password into this form field from a list 16:39:37 but seriously, it's for turning an intractible problem into a manageable problem 16:39:47 * Technicool glances at the clock 16:39:48 importing data from a survey service is an easy problem 16:39:51 ok, I can look for other services 16:39:54 semiosis: yup 16:40:12 surveyces... 16:40:14 #action jmw to look at services for farming out simple tasks 16:40:32 semiosis: can you look at survey services? 16:40:47 sure hit me with an #action 16:40:52 you do it! 16:40:55 heh 16:40:59 ah ok 16:41:12 #action semiosis to look at survey services not called surveymonkey 16:41:18 ha 16:41:20 #action semiosis to look at non Family Fued based survey services 16:41:27 ha 16:41:30 ok, I have to run 16:41:36 and I think Technicool does, too 16:41:43 * Technicool also has to run and/or walk fast 16:41:44 any burning issues we're missing? 16:41:49 going once 16:41:51 going twice 16:41:53 ... 16:41:54 the packages 16:41:59 Technicool: doh 16:42:04 older versions, you mean? 16:42:19 and non RHEL 16:42:33 we can pull the tarballs from git and then just rpmbuild via existing .spec files 16:42:36 I think 16:42:38 what ever happened to that debian packager req? 16:42:39 right? 16:42:52 Technicool: not sure. need ot ask quaid 16:42:53 Technicool: he's on vacation 16:42:55 for rhel, im not concerned since we have most of what we need 16:42:57 heh 16:43:03 yeah 16:43:13 but for the non rhel stuff, i think it looks bad to have gone this long without addressing it 16:43:19 Technicool: ok 16:43:33 Technicool: kkeithley set up a debian repo and i set up an ubuntu ppa, what's your issue specifically? 16:43:41 and, for things like hadoop plugin, but yes that should be a simple rpmbui.d 16:43:45 Technicool: since kkeithley did the builds, we should bribe him to do the others 16:43:55 semiosis: +1 16:43:56 johnmark, issues are 16:44:03 I think it's about getting older versions on the download server 16:44:08 1) is that available via download.gluster.com 16:44:11 right 16:44:28 2) do we know whether the packages were built with real debain expertise 16:44:37 in semiosis case i think we can rest assured 16:44:43 lol 16:44:46 yeah, and we can mirror his repo pretty easily 16:44:51 semiosis: I stifled my laughter ;) 16:44:53 it was a compliment... 16:45:05 appreciated :) 16:45:07 Technicool: but for the debian stuff, we had complaints the first day 16:45:09 and then none 16:45:17 whihc tells me that they're probably fine 16:45:19 technically i was able to build the deb packages, but i sure didn't rush out and say "here everyone!" 16:45:21 ;) 16:45:21 although it wouldn't hurt to sak 16:45:26 er ask 16:45:27 Or both debian users gave up. 16:45:33 lol@both 16:45:35 JoeJulian: also a possibility 16:45:43 ok, we need to ask 16:45:49 ask what? who? 16:45:51 well, i know of at least one very very very large debian customer 16:45:53 Technicool got it. 16:46:09 semiosis: send out a note to gluster-users saying "we have multiple debian version support now" 16:46:13 but since they are paying, they will probably get direct support on it 16:46:15 ...please let us know if they actually work 16:46:23 ok, that works for me 16:46:24 Technicool: ah, good point 16:46:29 ok? 16:46:30 cool 16:46:41 just want to make sure that non-RHEL users feel loved too 16:46:51 #action jmw to send out message to gluster-users informing everyone of new debian repos and asking for feedback 16:46:56 yup 16:46:58 Technicool: +1 16:47:01 ok.. cool 16:47:04 now I *really* have to go 16:47:06 ok, cool, gotta run 16:47:09 johnmark: See if any of those very very ... large paying customers want to sponsor some of those educational grant case studies. 16:47:15 johnmark: hold off on that email 16:47:19 semiosis: ok 16:47:20 lets chat before you do 16:47:23 semiosis: sure thing 16:47:26 but doesnt have to be now, we both have to go 16:47:30 but it will be quick 16:47:31 yup, no worries. ttyl 16:47:37 #endmeeting