15:02:28 #startmeeting Weekly GlusterFS Community Meeing 15:02:28 Meeting started Wed Aug 6 15:02:28 2014 UTC. The chair is JustinClift. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:02:28 Useful Commands: #action #agreed #halp #info #idea #link #topic. 15:02:45 Roll call 15:02:53 Who's here? 15:02:53 * kkeithley is humming along 15:02:54 yo es aqui 15:03:06 * ndevos is mostly here 15:03:10 * xavih is here 15:03:21 * bennyturns is here 15:04:26 Cool 15:04:52 #topic Agenda items from previous meeings 15:04:55 meetings 15:04:58 btw: https://public.pad.fsfe.org/p/gluster-community-meetings 15:05:17 "JustinClift to ask Pranith how bad the mem leak in 3.4 is that his patch fixes" 15:05:31 Done. It's not bad enough to be urgent 15:05:46 "JustinClift to ping Pranith about his mem leak patch for 3.4, and find out if it's in other branches as well" 15:06:34 Done. Hmmm, I don't rememeber the answer though. I think it was a case of nothing needing to be done for other branches, but I don't remember why. :/ 15:06:48 "JustinClift to ask on gluster-devel about our Planning3.6 schedule" 15:06:57 Not done, as someone else already asked 15:07:31 Hmmm, there are a few items for Pranith abt 3.6 test days 15:07:57 He's not here atm, so I guess we'll figure out the test day stuff with hagarth instead :) 15:08:19 "lalatenduM send a bug triage email announcing the process, and call for volunteers" 15:08:20 JustinClift: we need to start the test days, right now we are lagging behind the schedule on planning36 15:09:11 hagarth: Yeah. I thought that was ue to prioritisation on 3.5.x stuff. If we still need to do that, that's ok. We just need to communicate it 15:09:30 JustinClift: me and Pranith will send out communication on the revised plan tomorrow 15:09:41 hagarth: AI it 15:09:43 :) 15:10:00 hagarth: are the problems with 3.6 severe enough to not do an alpha release to test some features? 15:10:21 hagarth: Do you have time to respond to this? http://supercolony.gluster.org/pipermail/gluster-users/2014-August/018313.html 15:10:25 ndevos: Good idea 15:10:37 We can still have people test it out independantly 15:10:44 ndevos: not quite, we can definitely do an alpha release. I ran into a problem while running the release script from release-3.6. 15:11:04 hagarth: ah, ko 15:11:06 *ok 15:11:19 JustinClift: will do, lagging behind on the mail storm on gluster-* now :) 15:11:28 np ;) 15:11:32 how urgent is an alpha when people can test with the nightly build? 15:11:48 #action hagarth and pranith to communicate new plan for 3.6 tomorrow 15:12:22 kkeithley: Good question. It's just an expecation thing probably. An alpha gives a defined target. 15:12:40 I have some pending modifications to disperse (some on master and some on release-3.6) that would be good to have in 3.6 to avoid known problems 15:12:43 kkeithley: hmm, I think there is no 3.6 nightly build... I was going to check that 15:12:52 We can definitely encourage ppl to test the nightlies 15:12:59 Heh 15:13:03 xavih: right, will merge them before the release. 15:13:12 hagarth: ok :) 15:13:13 ndevos: AI yourself for checking that? 15:14:10 JustinClift: can a non-chair do that? 15:14:10 Anyone know the status of "lalatenduM send a bug triage email announcing the process, and call for volunteers" ? 15:14:16 ndevos: Yep 15:14:27 #action ndevos to check on 3.6 nightly builds 15:14:45 I'll ping lala and ask I guss 15:14:48 guess 15:14:59 lalatenduM, ^^^^ 15:15:00 :) 15:15:07 #action JustinClift to ask lala about status of "lalatenduM send a bug triage email announcing the process, and call for volunteers" 15:15:23 "prashanth to solicit feedback on who would join the weekly meeting when it takes place 2 or 3 hours earlier" 15:15:37 Done. Large amount of votes (66 in total) 15:15:50 Pretty clear majority for the 12:00 UTC time 15:16:00 So, we'll do that next week and see how things go 15:16:11 JustinClift, Will do this week , sorry for the delay 15:16:41 Prashanth mentioned there could have been cheating, but really, there may not have been and it doesn't seriously hurt us to try it out 15:17:06 So, from next week we'll do the meetings at 12:00 UTC 15:17:22 JustinClift: cool 15:17:26 lalatenduM: Tx. Please update the etherpad abt that: https://public.pad.fsfe.org/p/gluster-community-meetings 15:17:36 JustinClift, will do 15:17:58 "JustinClift to get initial GlusterFS Consultants and Support Companies page online" 15:18:13 Working on it. Had some technical problems with Middleman, now resolved 15:18:20 So, no blockers any more :) 15:18:25 Moving on... 15:18:29 #topic 3.4 15:18:41 3.4.5 seems pretty decent for people 15:18:49 1 mem leak path (non-urgent) 15:18:59 1 crashing bug (under investigation) 15:19:05 Nothing requiring panic :) 15:19:25 JustinClift: what is the mem leak problem? the one reported by JoeJulian? 15:19:25 worth considering a 3.4.5-2 "patch" release? 15:19:47 I think we should wait for rca of crash bug 15:19:51 kkeithley: Lets see what the crashing bug patches turns out like 15:19:56 yeah 15:20:03 agreed 15:20:22 hagarth: Not sure. Pranith did it not long ago, but I don't rememeber the associated issue 15:20:31 hagarth: Didn't seem urgent 15:20:34 JustinClift: I think JoeJulian's memory leak problem should be addressed too. He reported it over the last day or so. 15:20:42 Oh 15:20:53 That's new to me 15:21:00 there are a few critical rebalance fixes that have got into master 15:21:08 Is there a BZ for it? 15:21:14 Applicable for 3.4? 15:21:18 we would need all those fixes to be backported to all releases 15:21:21 kkeithley: checking 15:21:47 kkeithley: https://bugzilla.redhat.com/show_bug.cgi?id=1127140 15:21:48 Bug 1127140: unspecified, unspecified, ---, gluster-bugs, NEW , memory leak 15:22:38 hagarth: Can you get someone assigned to it, so it gets figured out/fixed? 15:22:58 JustinClift: will do. we also need the rebalance fixes to be backported. 15:23:24 someone recently added 1123289 to 3.4.6 tracker 15:23:25 k, so this stuff all seems 3.4.6 worthy. We just don't have a timeframe yet 15:23:57 Lets figure out the solutions and urgency of this stuff, and then plot & plan the next release around that 15:24:22 hagarth: Can you AI those bits, so we follow up next week? 15:24:50 #action kkeithley & hagarth to follow up on fixes for 3.4.6 15:24:58 :) 15:25:01 Moving on... 15:25:05 #topic 3.5 15:25:14 finish 3.4? 15:25:22 Oops 15:25:30 Sorry 15:25:37 ;-) np 15:25:39 #topic 3.4 attempt 2 15:25:57 "https://bugzilla.redhat.com/show_bug.cgi?id=1127140" 15:25:58 Bug 1127140: unspecified, unspecified, ---, gluster-bugs, NEW , memory leak 15:26:00 Gah 15:26:03 just to check, 3.4 is also supported during the 3.6 release, and support ends with 3.7, right? 15:26:05 1 sec 15:26:14 ndevos: right 15:26:22 "Should the patch/fix backport request page just point to the respective tracker BZs?" 15:26:24 okay :) 15:26:38 mkay 15:26:42 JustinClift: I think it does that? 15:26:53 This sounds like a decent idea to me, as it's one less place for ppl to look 15:26:58 does it? It didn't use to 15:27:04 ndevos, hagarth shouldnt it be , as soon as 3.6 is released ? 15:27:21 "3.4 EOL when 3.6 GA?" 15:27:25 * JustinClift catches up 15:27:26 yep :) 15:27:44 hchiramm: I don't think we should do that 15:27:54 kkeithley: Yeah, I don't think it does yet. But it sounds like a decent idea to 15:27:58 For the EOL bit... 15:28:15 I would prefer 3.4.x to be stable and supported for a while 15:28:22 We still don't have a 3.5 which is super great yet 15:28:30 Yeah, me too 15:28:50 kkeithley: http://www.gluster.org/community/documentation/index.php/Backport_Wishlist points to the tracker for 3.5 :D 15:28:54 For EOL, IIRC we used to only support two versions. But we have more people doing releases now, so it's not unreasonable 15:29:01 So, I'd rather we don't drop 3.4 offical supported status until we're happy with 3.6 stability 15:29:07 hagarth, if we move 3.4 to EOL with 3.6 release, its indirectly pushing users to upgrade and make latest versions stable.. 15:29:09 thats my view 15:29:14 ndevos: okay, I haven't looked in a while 15:29:18 hchiramm: Yes 15:29:30 still we see lots of users sticking with older versions and keep reporting the issues on those.. 15:29:45 JustinClift, you mean s/we're happy with 3.6 stability/we're happy with 3.5 stability/ right? 15:29:54 hchiramm: sure, we can be selective about what we want to fix in 3.4.x 15:29:57 hchiramm: For data though, there is massive downside of pushing users to not-stable-enough versions too early 15:30:19 lalatenduM: So we're happy with _both_ 3.5 and 3.6 15:30:28 but stability of our latest versions get delayed :) 15:30:29 Assuming 3.5 stablity won't be too far off 15:30:38 did you mean 3.4, 3.5, and 3.6? 15:30:46 hchiramm: Sure. But people not losing data is _way_ more important 15:30:58 kkeithley: Now I'm confused 15:31:05 by the time we release 3.6 , our 3.5 should be stable.. thats the point :) 15:31:09 JustinClift, LOL 15:31:12 Heh 15:31:31 JustinClift: oh, I don't know what you meant, I guess, by "So we're happy with _both_ 3.5 and 3.6" 15:31:36 hchiramm: yes, once 3.6 is released.. there will be very few releases off 3.4.x 15:31:48 hchiramm: Agree. At the same point, lets not burn people 15:32:13 agree with hagarth 15:32:14 hchiramm: upgrading a storage infrastructure twice in a year is not fun :) 15:32:48 Anyway, we'll have a clearer picture when we see 3.6 closer to release 15:33:01 I thought they should take risk with atleast one version.. :) 15:33:09 It could turn out to be magical and full of ponies and unicorns, and we might have everyone upgrade :) 15:33:13 to be clear then, we 3.4 will not EOL when 3.6 GAs 15:33:13 any way, I was just sharing my view. 15:33:18 no objections. :) 15:33:21 hchiramm, why someone will take risk with storage :) 15:33:22 hchiramm: definitely not with storage 15:33:24 hchiramm: np :) 15:33:31 kkeithley: Yep 15:33:56 Now, moving on... 15:34:04 #topic 3.5 again :) 15:34:18 3.5.2 is out (yay) 15:34:33 But still getting reports of serious memory leaks with it (not-yay) 15:34:54 Lets try and get a 3.5.2-2 out asap maybe, if they can be tracked own? 15:34:55 down 15:34:56 JustinClift: what are the reported leaks? 15:35:02 1 sec, it's on list 15:35:07 blockers for 3.5.3 - https://bugzilla.redhat.com/showdependencytree.cgi?maxdepth=1&hide_resolved=1&id=glusterfs-3.5.3 15:35:07 * JustinClift gets the emails 15:35:40 not sure if there is a memleak on that list... 15:36:32 can't find any obvious leak report in there 15:36:53 http://supercolony.gluster.org/pipermail/gluster-users/2014-July/018218.html 15:37:10 * JustinClift really doesn't like the archives splitting things per month 15:37:16 JustinClift: did we convert that to a bz? 15:37:25 hagarth: There's more on the August page too 15:37:33 hagarth: No idea 15:37:44 I just noticed the thread today 15:37:53 JustinClift: that email(thread) does not list a bug# 15:37:56 But, there was someone else saying the same thing 15:38:03 ndevos: k, we probably didn't then 15:38:17 we should encourage the OP to log a bug 15:38:31 Or Poornima G 15:39:01 #action JustinClift to encourage Tamas Papp to log a bug bout http://supercolony.gluster.org/pipermail/gluster-users/2014-July/018219.html 15:39:32 It sounds like a guys are looking into it, so I'll check the thread out properly and see about getting it BZ'd 15:40:11 ndevos: With the 3.5.3 tracker, do you reckon it's something we can spit out a 3.5.3 in say 2 weeks? 15:41:28 JustinClift: probably, but the bugs in POST require the patches in the master branch, mostly these are pending review/merging 15:41:45 Ahhh gotcha 15:42:32 ndevos: Do you have time to send a email to gluster-devel, something like "We need these patches in master reviewed, so the backports can be merged in 3.5.3"? 15:42:33 I'm also not sure how critucal those bugs are, maybe we should wait for a bug+fix for that memleak 15:42:40 ndevos: Sure 15:42:47 Lets try and get it all in the same thing :) 15:43:19 JustinClift: sure, I can check that 15:43:34 ndevos: k, Make an AI for yourself so we check next week 15:43:49 #topic 3.6 15:44:06 #action ndevos will verify status of patches proposed for 3.5.3 and request reviews for those patches in master 15:44:37 So, we're a bit behind schedule, and hagarth + Pranith will send out the updated schedule probl tomorrow 15:44:42 for 3.6 that is 15:44:48 JustinClift: right 15:44:54 So, doesn't sound like anything else for this item atm? 15:45:19 Moving on 15:45:25 #topic Other Agenda Items 15:45:28 "cppcheck fixes nag. We missed 3.5.2 because they have not been accepted on master, and now 3.6. We got them in 3.4 so that Ubuntu would update to a recent release. This is sad." 15:45:41 * kkeithley is especially sad 15:45:42 k, whoever's that is, you have the er... floor :) 15:45:45 +1 sad indeed 15:46:02 kkeithley: They're needing reviews? 15:46:10 right, let us expedite our reviews for that 15:46:11 yes, reviews are need 15:46:14 ed 15:46:30 yup, these fixes have stayed in review for long now 15:46:56 k. Can we do an _assign_ of reviews maybe, so they get prioritised? 15:47:16 I'm just going to keep being a squeaky wheel 15:47:21 eg "Dear foo, it's your job to get these reviewed: cr 1111, 222" 15:47:22 k 15:47:33 hagarth: Hows that leader board looking? 15:47:37 hagarth: Updated? 15:47:57 kkeithley, I think a mail in gluster-devel will elp 15:48:02 help* 15:48:02 hagarth: Also, have we put any work into tying real world rewards to it yet? 15:48:03 JustinClift: have an offline copy, will post it online tomorrow. 15:48:17 hagarth: Just thinking that might help get this fixed :) 15:48:32 JustinClift: I think we will get there eventually :) 15:48:36 ;) 15:48:47 let us focus on getting the leaderboard updated regularly now 15:49:05 kkeithley: Are you ok to squeaky wheel to gluster-devel about the reviews needed for master for it? 15:49:06 I we should publish them as blogs 15:49:22 lalatenduM: That's a good idea 15:49:39 hagarth: can I RFE to get mailinglist stats in the leaderboard as well? 15:50:09 ndevos: bitergia already does that for us 15:50:31 ndevos: #link http://bitergia.com/projects/redhat-glusterfs-dashboard/browser/ 15:50:35 hagarth: oh, nice! Link in the next stats email/blogpost? 15:50:42 or that :_ 15:50:45 :) 15:50:48 If we do blog posts for it, we might as well pull the appropriate bitergia stats into it so people don't have to go to an extra place 15:50:50 sure, will do 15:51:23 * JustinClift likes the blog idea for this 15:51:38 Good visibility, plus auto-tweet about it from doing that 15:51:45 JustinClift: the blog is all yours ;) 15:52:09 hagarth: k, forward the email bits to me, or put them on gluster-devel or something, and I'll make a blog post out of it 15:52:32 #action JustinClift to create blog post of latest leaderboard stats, after receiving from hagarth 15:52:34 JustinClift: will do 15:53:02 #action kkeithley to email gluster-devel about the reviews needed for cppcheck stuff on master 15:53:03 JustinClift: are you the new community manager? 15:53:09 ndevos: No 15:53:23 :) 15:53:32 there's a new community manager who will start in a couple weeks 15:53:41 JohnMark's replacement 15:53:42 It's just no-one's doing this stuff with a focus atm, so I'm trying to fill that gap until new dude or dudette is in place :) 15:54:13 Moving on... 15:54:15 "packaging 3.5.2 for Debian" 15:54:25 Go for it whoever :) 15:54:42 We've set up a Wheezy box and will start building our own packages 15:54:49 instead of waiting around 15:54:51 JustinClift: I thought pmathai uploaded 3.5.2 debs to unstable ? 15:54:56 We being me and hchiramm 15:55:10 kkeithley: Cool. 15:55:13 nice! 15:55:16 unstable? 15:55:24 for download.gluster.org 15:55:30 3.5.1 bits are there. No 3.5.2 15:55:30 kkeithley: ah ok 15:55:43 No 3.5.2 yet 15:55:50 Is the wheezy box in RH firewall, or publicly accessible? 15:56:00 https://packages.debian.org/sid/glusterfs-client 15:56:02 It's beind the firewall 15:56:04 * JustinClift is wondering if we can get Community members to help out 15:56:07 Ahhh, k 15:56:26 Well, at least it'll be on our own schedule 15:56:37 I did it once, about a year ago, in a vm on my desktop. Now we have a vm in the lab 15:56:38 JustinClift, we can't have a debian vm in rackspace? 15:56:42 sid seems to have 3.5.2 in unstable 15:56:42 and more people to share the load 15:56:46 hagarth, oh.. then somehow we could pull it to download.gluster.org 15:56:49 kkeithley: If you want a box on public network, we can setup a rackspace vm very easily 15:57:05 lalatenduM: Yeah 15:57:18 hchiramm: right, let us sync up with Patrick Matthai and offer him any help for avoiding delays 15:57:23 It just needs to be somewhere besides a vm on my desktop 15:57:33 sure. 15:57:37 It's definitely possible to ask on the public mailing lists if anyone's interested in helping us build official deb's. Patrick Matthai, etc 15:57:46 kkeithley: Cool. :) 15:57:56 machine is up, I'll consult my notes from last year. Might need some hints from pmatthai or semiosis 15:58:03 kkeithley: You have Rackspace access. Want to spin up a 512MB or 1GB VM there? 15:58:20 I can do it if you want me to, but I'll just hand it off to you once it's running 15:58:25 I've got a machine here in the lab already running 15:58:31 k. 15:58:37 JustinClift, I think we should let kkeithley figure this out , then think abt moving it to public 15:58:46 Yeah, better idea :) 15:58:52 and additional packages 15:58:56 * JustinClift butts out :) 15:59:07 The package set we've got is suffering from some bitrot 15:59:26 e.g. no -dev and no -geo-rep 15:59:51 Gotcha 16:00:11 Ask on gluster-devel and gluster-users if ppl have suggestions for what else? 16:00:14 not sure what else we ought to have 16:00:53 Oops, we've hit the 1 hr meeting time 16:01:14 The other two are just plugs/reminders of some new things 16:01:18 Lets finish the other items wquitckly 16:01:22 Yeah, go for it 16:01:47 People looking for easy bugs to fix should look at http://download.gluster.org/pub/gluster/glusterfs/static-analysis 16:02:04 Lets blog about it again 16:02:08 I'll be adding more as time permits 16:02:09 For extra attention 16:02:10 kkeithley: we probably should find a permanent place for these on gluster.org too 16:02:12 yup 16:02:20 hagarth: Good idea 16:02:23 kkeithley: are those bugs listed in https://bugzilla.redhat.com/query.cgi?product=GlusterFS&keyword=EasyFix&status=NEW ? 16:02:28 JustinClift, kkeithley I have updated the lnks in respective wiki page 16:02:36 lalatenduM: Tx :) 16:02:53 lalatenduM: That's for this static analysis stuff or ? 16:03:01 ndevos: no, I haven't automated that part yet 16:03:09 JustinClift, yeah static analysis stuff 16:03:12 :) 16:03:49 k, obviously integrating the static analysis bits with the EasyFix BZ tags would be useful 16:03:59 and let's roll the bugzilla sub-components to next week's meeting 16:04:05 yep 16:04:09 uh, that should be https://bugzilla.redhat.com/buglist.cgi?product=GlusterFS&bug_status=NEW&keyword=EasyFix 16:04:19 k, thanks for the meeting everyone :) 16:04:22 #endmeeting