12:00:30 #startmeeting Weekly community meeting 8/Jun/2016 12:00:30 Meeting started Wed Jun 8 12:00:30 2016 UTC. The chair is kshlm. Information about MeetBot at http://wiki.debian.org/MeetBot. 12:00:30 Useful Commands: #action #agreed #halp #info #idea #link #topic. 12:00:30 The meeting name has been set to 'weekly_community_meeting_8/jun/2016' 12:00:35 Hi all! 12:00:45 kshlm: Hi 12:00:54 dobry den 12:00:57 aravindavk, Hi! 12:00:58 Welcome back kshlm 12:01:06 :) 12:01:09 namaskara 12:01:14 :) 12:01:31 * ndevos is here, but also having breakfast in a hotel... 12:01:42 #topic Rollcall 12:01:48 o/ 12:01:48 * kotreshhr is here 12:02:01 * atinm is in 12:02:03 * poornimag is here 12:02:06 * rastar is here 12:02:07 :\ the agenda isn't prepared. 12:02:19 Give me a couple of minutes. 12:03:19 * anoopcs is here 12:03:23 * rafi is here 12:03:28 * partner here 12:03:55 * msvbhat_ is here 12:04:05 Hey msvbhat_ ! 12:04:09 Okay! 12:04:14 Let's start 12:04:18 Hello... :) 12:04:39 I'll host the next meeting, so I'm skipping the host selection 12:04:47 #topci GlusterFS-4.0 12:04:55 #topic GlusterFS-4.0 12:05:06 Time for some updates! 12:05:22 I've seen few patches coming from DHT2, so that's a good progress :) 12:05:56 I'm getting back to GD2 again after my break. 12:06:11 I don't have much updates from jdarcy on NSR side as I missed out a meeting with him 12:06:25 I'll be starting to extend the trasaction framework to work for multi-nodes. 12:06:41 Anyone else working on 4.0 here? 12:07:08 I don't see jdarcy overclk or shyam 12:07:18 So I guess I'll move to the next topic 12:07:20 any volunteers for Brick Multiplexing? We need people to work on this part as well :) 12:07:57 kshlm: shyam posted the skeleton patches for posix and dht2-server side xlator! 12:08:04 I'd love to, but I've got things full already 12:08:13 Sorry I'm late. 12:08:20 kotreshhr, Cool! thanks. 12:08:22 just in time 12:08:27 jdarcy, no issues, your turn for NSR updates 12:08:28 jdarcy, Not too late. 12:08:31 jdarcy, :) 12:08:56 Sadly, not much to update. I've been busy preparing for the SA Summit (where I am now) and Red Hat Summit. 12:09:19 Oh, and talking trash about competing projects. ;) 12:10:09 jdarcy, I can sense that :D 12:10:48 They won't be competing any time soon. 12:11:40 Thanks for the updates jdarcy 12:11:46 #topic GlusterFS-3.8 12:12:14 ndevos, jiffin, Would you please share your updates? 12:12:18 some more progress was made the last weeks 12:12:22 more patches got merged 12:12:30 Any dates for the release yet? 12:12:39 but, as I'm travelling this week, I wont be able to do the release now 12:12:48 do we need mount option to set glusterfsd process niceness like i've already done for OOM score? 12:12:54 and, due to the Gerrit update, I was not able to releease anything earlier 12:13:20 post-factum: 'need' is flexible :) 12:13:51 I haven't been able to work on the socket fixes for encrypted and IPv6 connections yet. I would still like to get them in, but not sure if I can. 12:13:54 release date for 3.8 is now planned tuesday/wednesday next week 12:14:10 a concern here, wrt 3.8 and generally as well..., getting reviewer time is getting harder and harder 12:14:11 so we're holding patches for release-3.8 branch for now? 12:14:17 that means all changes need to be merged before Monday, LATEST 12:14:35 agree with poornimag 12:14:51 and I don't see that happening with a broken build system 12:15:09 not really holding patches, but changes should be carefully considered before merging, no risky things 12:15:26 spalai, poornimag : could you send the list of patches to the maintainers list which need urgent attention? 12:15:48 can we have something like review backlog for patches those were not reviewed for certain amount of time 12:15:50 atinm, beein doing it 12:15:53 atinm: in general 12:16:05 indeed, any changes that REALLY REALLY REALLY need to get in 3.8.0 should be mentioned in emails to the maintainers list 12:16:44 #link https://bugzilla.redhat.com/showdependencytree.cgi?id=glusterfs-3.8.0&hide_resolved=1 12:17:19 I don't like emails with subject review needed with just a review link, if the mail explains why that urgency and a little background on the problem it helps maintainers to prioritize them, having said that I am not saying that maintainers shouldn't be actively checking the backlogs but it has to be a collaborative approach 12:17:19 only very few bugs in that list are not in MODIFIED or ON_QA state yet, add bugs that require changes in 3.8.0 12:17:28 ndevos: can we do something on getting reviewers attention (not just for 3.8 in general) 12:17:57 spalai, We can. 12:18:20 spalai: there are many maintainers that should be responsive for doing reviews, send an email (format like atin just mentioned) to the list 12:18:45 One of the suggested ideas for updating our release timelines, is to have merge windows. This would be the time for reviews. 12:19:11 personally I do not click links to reviews, if the email does not contain a useful message/topic, I'm not reviewing patches for all components 12:19:34 and also the option of peer reviews to be considered as well 12:20:19 a maintainer would always have more confidence on a patch if it has been reviewed by someone else who works on that same component 12:20:23 developers working on components should watch out for patches, and do reviews on those too 12:20:45 FWIW, here's my review "dashboard" - http://review.gluster.org/#/q/status:open+project:glusterfs+branch:master+-label:CentOS-regression%253C0+-label:NetBSD-regression%253C0+-label:Code-Review%253C0 12:21:01 it is easy to setup notifications, see http://gluster.readthedocs.io/en/latest/Contributors-Guide/Guidelines-For-Maintainers/ for some examples 12:21:50 I'll generally look at anything that already has a +1 from someone else (e.g. a team lead) first. 12:21:55 ndevos: thanks for the link. 12:22:11 atinm, ndevos there have been several mails that have been sent, i can resend those for reference, mail is not good enough for tracking it.., bugging for months for review is not a good position to be in 12:22:21 anyway, 3.8.0 was delayed for more than 6 months now, and we really should release soon, I dont plan on delaying it more than until next week 12:22:27 ndevos, oh, ok i shall look at it 12:23:06 ndevos, I agree. 12:23:46 so, only the most urgent blocking patches need to get merged, others will get done for 3.8.1 a month later 12:23:57 poornimag, so if you attach the release tracker to the bug corresponding to the patch it does your statement hold true (waiting for months) ? 12:25:07 for changes backported to 3.8, I plan to enforce http://thread.gmane.org/gmane.comp.file-systems.gluster.maintainers/694 where possiblie 12:25:14 atinm, the bugs are added to the tracker 12:25:16 Aren't we moving to the newer release schedules once 3.8 is released? 12:25:34 that describes the requirements etc of patches that may get backported, and which changes are banned 12:25:36 ie. new releases every 3 months, with a LTS release every year? 12:26:24 kshlm: with that new scheme, I still assumed we would do monthly bugfix releases for the active major releases 12:26:34 * kshlm wonders what happened to the release discussion. 12:26:49 poornimag, that means the release can't go in without these patches 12:27:35 ndevos, That's an interesting mail. 12:27:38 poornimag: well, the tracker is to give some idea of what is still needed, but we can move bugs to a new 3.8.1 tracker too ;-) 12:28:30 kshlm: the intention is to get only bugfixes backported (except for experimental features in the release, they have more options) 12:28:34 ha ok, but i guess it will be the same thing for next release 12:28:41 ndevos, we should write up such a rule set for accepting patches into minor releases. 12:28:50 And enforce it obviously. 12:29:05 kshlm: yes, the email is a draft for it, but none of the maintainers replied :-( 12:29:14 I will once I've read it. 12:29:26 I've had a huge backlog of email to get through. 12:29:27 great, thanks! 12:29:33 sure, no worries 12:29:49 * ndevos suggests to move to the next topic 12:29:52 Okay. 12:29:57 I was thinking the same. 12:30:07 #topic GlusterFS-3.7 12:30:21 ndevos, I think majority of the maintainers already replied, we are waiting on hagarth to send an email on this 12:30:36 * post-factum is hunting memory leak in 3.7.11 again 12:30:45 I've seen 3.7.12 RCs, but have no idea when the release is planned. 12:30:57 Is anyone near hagarth? 12:31:03 #link https://www.gluster.org/pipermail/gluster-devel/2016-June/049767.html 12:31:23 our memory leak issues dropped ~10x when moving from 3.6.9 to 3.7.11 but testings continue (libgfapi, openstack,..) 12:31:39 almost non-existing with 3.8 rc2 12:31:41 georep patches blocked due to regression issues 12:31:42 kshlm, maintainers have to ack the build 12:31:50 post-factum, I still keep thinking some crazy thread issues are happening. 12:31:53 partner: that's great news 12:31:56 kshlm, hagarth expects that to be done by this week 12:32:12 kshlm: i need extra info on how i could help in debugging that 12:32:21 But I'm just looking at it from behind very biased glasses right now. 12:32:39 atinm, Yeah. That's something I should do as well. 12:33:02 atinm, Do you have a link to hagarth communication regarding that? 12:34:40 atinm: hmm, I'm wondering which maintainers replied, I do not find any responses (this quickly), and the gmane archove also does not have them? 12:35:02 kshlm, looking for it 12:35:12 ndevos, give me couple of minutes to get to that thread 12:35:18 I seem to have lost emails to the maintainers list. I only see the regression-burn in mails. 12:35:51 atinm: take your time :) 12:35:55 I don't see either ndevos mail on backport rules or hagarth discussion on 3.7.12 12:36:55 kshlm, http://www.gluster.org/pipermail/maintainers/2016-June/000847.html 12:37:17 atinm, Thanks. 12:37:30 My seach-fu is lacking this week. 12:37:39 #link http://www.gluster.org/pipermail/maintainers/2016-June/000847.html 12:38:19 I'll work with atinm and get the ack for GlusterD before the end of the week. 12:38:41 #action kshlm/atinm to ack 3.7.12 before the end of the week 12:38:58 Other maintainers, you need to provide your acks! 12:39:20 Okay. 12:39:27 #optic GlusterFS-3.6 12:39:36 #topic GlusterFS-3.6 12:39:52 optic makes sense too 12:40:13 ndevos, http://www.gluster.org/pipermail/gluster-devel/2016-May/thread.html#49597 12:40:16 But zodbot doesn't understand. 12:42:29 atinm: uh, sure it is the right link? for the backport requirements? 12:43:04 * ndevos needs to find the event room, things start in a few minutes 12:43:15 thanks kshlm! 12:43:31 ndevos, see ya later 12:43:31 ndevos, doesn't the heading talk about "alternate release process" ? 12:43:51 Anyone want to talk about 3.6? 12:44:02 I'll skip if we don't have anything. 12:44:29 atinm, The link you gave is to new proposed release timelines. 12:46:08 No 3.6 updates this week. 12:46:09 I believe that's what I was looking for 12:46:15 LTS and non LTS stuffs 12:46:28 Onto the AIs from last week 12:46:37 #topic Last weeks AIs 12:46:43 I don't think any of them is done. 12:46:57 So I'll just carry them forward. 12:47:27 #topic Open floor 12:47:41 misc, are you around? 12:47:43 oh, you don't want status on NFS-Ganesha or Samba? 12:47:52 kkeithley, Ah! 12:47:55 Forgot that! 12:47:59 misc, do we have any luck on build system? 12:48:04 I've not been used to the new agenda yet. 12:48:18 #topic Ganesha 12:48:24 kkeithley, You're up! 12:48:43 nfs-ganesha continues slow march toward 2.4 GA. DanG has been running valgrind on next (2.4) branch. Found some things that will apply to 2.3.x 12:48:49 atinm: what do you mean ? 12:49:32 misc, I believe nigelb was working with you on the regression failures what we have been noticing since last couple of days 12:49:36 kkeithley, And anything new about ganesha+gluster? 12:49:45 atinm: we had a few of them 12:49:53 misc, I am talking about that as we still see the regressions are not getting through 12:49:57 and i also had to handle a few emergencies for others projects 12:50:15 so speaking of "the problems" is not really helping me to answer 12:50:25 nothing specific for ganesha+gluster, no 12:50:27 we fixed the issue regarding crash of nss 12:50:33 Thanks kkeithley 12:50:37 I looked at the netbsd issue with builder full 12:51:00 Is anyone around for Samba updates? 12:51:07 and I am now hardening the servers since 1 of them participated in a ddos yesterday 12:51:08 misc, are you guys aware of that the job is picking a different patch all together? 12:51:22 atinm: yes, cf gluster-infra 12:51:27 but I didn't look much at that 12:51:57 misc, ok 12:52:00 I'll skip Samba updates for today. 12:52:06 #topic Open floor 12:52:37 I think atinms discussion with build issues is done. 12:52:58 To add a little more information to what's happening. 12:53:22 Q: it's not clear to me. Which version is going to be the LTS? 3.7? 12:53:34 What I've observed is that for some reason Jenkins is voting for very old jobs, on new patches. 12:53:43 Still haven't figured out why this is happening. 12:53:54 kkeithley, I believe it's going to be 3.8 12:54:29 kshlm, and it's the other way around as well, its picking patches on adhoc basis, no? 12:54:42 atinm, Not what I've seen. 12:55:01 Do you have an example? 12:56:04 kshlm, ok so what you are saying may be correct, when I open a patch and go to the regression link I see a different patch there (which is a old job), so basically your statement is correct 12:56:22 atinm, That is what is happening 12:56:53 I'm now wondering if its possible that the old jenkins server has come back from the dead. 12:56:56 kshlm, ok, thanks for the correction, I misunderstood it 12:57:38 kkeithley, Are you happy with the answer? 12:57:47 We're nearly outta time. 12:59:12 I think kkeithley's happy. 12:59:30 Thanks every one for attending today's meeting! 12:59:34 And remember 12:59:44 If you're attending any event/conference please add the event and yourselves to Gluster attendance of events: https://public.pad.fsfe.org/p/gluster-events 12:59:44 Put (even minor) interesting topics on https://public.pad.fsfe.org/p/gluster-weekly-news 12:59:45 Use the following etherpad for backport requests  https://public.pad.fsfe.org/p/gluster-backport-requests 12:59:56 Goodbye! 12:59:59 #endmeeting