12:03:29 <hagarth_> #startmeeting Gluster Community Weekly 12:03:29 <zodbot> Meeting started Wed Nov 25 12:03:29 2015 UTC. The chair is hagarth_. Information about MeetBot at http://wiki.debian.org/MeetBot. 12:03:29 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic. 12:03:29 <zodbot> The meeting name has been set to 'gluster_community_weekly' 12:03:49 <hagarth_> #chair hagarth_ kshlm 12:03:49 <zodbot> Current chairs: hagarth_ kshlm 12:03:59 <hagarth_> #topic Rollcall 12:04:04 <hagarth_> who do we have here today? 12:04:32 * msvbhat is here 12:04:37 * kshlm is here 12:04:43 * jdarcy ! 12:04:44 * raghu is here 12:04:53 * poornimag is here 12:05:04 <hagarth_> let's get going 12:05:17 <hagarth_> #topic AIs from last week 12:05:33 <hagarth_> ndevos send out a reminder to the maintainers about more actively enforcing backports of bugfixes 12:05:49 <hagarth_> I don't think this has happened and ndevos is away. So carrying on this AI. 12:05:53 <hagarth_> #action ndevos send out a reminder to the maintainers about more actively enforcing backports of bugfixes 12:06:03 <hagarth_> next AI - raghu to call for volunteers and help from maintainers for doing backports listed by rwareing to 3.6.7 12:06:10 <hagarth_> raghu: any updates on this? 12:06:38 * kkeithley_ is late 12:06:44 <raghu> hagarth_: I will do it for 3.6.8. 3.6.7 I have done today 12:06:55 <hagarth_> ok, modifying the AI 12:07:13 <hagarth_> #action raghu to call for volunteers and help from maintainers for doing backports listed by rwareing to 3.6.8 12:07:16 <kshlm> raghu, will you even get time to do it for 3.6.8? 12:07:34 <kshlm> Considering our release timelines and your travels. 12:08:13 <hagarth_> kshlm: I think it is a call for action.. raghu doesn't have to do it all by himself 12:08:19 <raghu> kshlm: Well, I can ask for maintainers. That mail I am going to send today or tomorrow. 12:08:30 * Humble is here 12:08:36 <raghu> oops not maitainers, volunteers 12:08:55 <kshlm> raghu, okay. 12:08:55 <hagarth_> raghu: noted, ok. 12:09:03 <hagarth_> next AI - hagarth to post a tracking page on gluster.org for 3.8 by next week's meeting 12:09:27 <hagarth_> this is done. though more details need to appear in the tracking page - https://www.gluster.org/community/roadmap/3.8/ 12:09:34 * rastar is here 12:09:54 * atinm arrives late 12:10:06 <kshlm> hagarth_, awesome page. 12:10:12 * anoopcs is present 12:10:18 <kshlm> How do we add stuff to it? 12:10:24 <kshlm> The glusterweb repo? 12:10:42 <hagarth_> kshlm: right, through a patch for glusterweb.git 12:11:12 * skoduri arrives late 12:11:15 <hagarth_> we will be discussing more about 3.8 later 12:11:21 <kshlm> hagarth_, okay. 12:11:35 <hagarth_> next AI - rafi1 to setup a doodle poll for bug triage meeting 12:12:06 <hagarth_> do we need this AI? looks like bug triage meetings are happening 12:12:17 <kshlm> I don't have context on this. 12:12:29 <kshlm> rafi, can you answer this? 12:12:52 <rafi> kshlm: I haven't done that yet 12:13:04 <kshlm> rafi, is it still required? 12:13:36 <kkeithley_> do we think turnout would be better at another time? Is the current time what's stopping people from attending? 12:13:36 <rafi> kshlm: I will discuss this in next bug triage meeting and get back to you 12:13:50 <hagarth_> rafi: ok, we will carry on this AI till then 12:13:58 <rafi> hagarth_: sure 12:14:10 <kshlm> rafi, kkeithley_, thanks. 12:14:13 <hagarth_> #action rafi to setup a doodle poll for bug triage meeting 12:14:23 <hagarth_> next AI - rastar and msvbhat to publish a test exit criterion for major/minor releases on gluster.org 12:14:35 <msvbhat> Done 12:14:37 <hagarth_> I just saw a related email on gluster-devel 12:14:51 <rastar> That is a preliminary mail 12:14:52 <hagarth_> msvbhat: it still needs to appear on a gluster.org page I think 12:14:58 <rastar> it just presents current state 12:15:03 <hagarth_> rastar: right.. 12:15:14 <rastar> we had other discussions on the exit criterion 12:15:18 <kshlm> rastar, In the nick of time :) 12:15:28 <rastar> We will be sending them over the week 12:15:37 <hagarth_> rastar, msvbhat: let us continue to track this AI? 12:15:41 <rastar> yes 12:15:46 <msvbhat> hagarth_: I can send a patch once its discussed 12:15:50 <hagarth_> ok cool. 12:15:54 <hagarth_> #action rastar and msvbhat to publish a test exit criterion for major/minor releases on gluster.org 12:15:57 <rastar> kshlm: :) 12:16:04 <hagarth_> next AI - jdarcy to send monthly update for NSR 12:16:05 <msvbhat> hagarth_: Patch to apear in gluster.org 12:16:13 <hagarth_> msvbhat: cool, thanks! 12:16:29 <jdarcy> Will do this later today. I realized there was one other piece of info I needed to collect. 12:16:50 <hagarth_> jdarcy: cool, thanks. 12:17:03 <hagarth_> next AI - samikshan to send status on Gluster Eventing 12:17:16 <hagarth_> I don't think samikshan is here, so carrying forward this AI 12:17:22 <hagarth_> #action samikshan to send status on Gluster Eventing 12:17:23 <atinm> hagarth_, it will be delayed, samikshan is attending some medical emergencies 12:17:34 <hagarth_> atinm: ah ok, thanks for the update. 12:17:58 <hagarth_> next AI - atinm/kshlm to have the GlusterD 2.0 design doc by end of next week 12:18:12 <kshlm> It's not the end of next week yet. 12:18:42 <atinm> kshlm is almost there to put it up I believe 12:18:50 <hagarth_> kshlm: so carrying forward this AI 12:18:58 <hagarth_> #action atinm/kshlm to put up the GlusterD 2.0 design doc by end of next week 12:19:01 <kshlm> hagarth_, yup 12:19:17 <hagarth_> next AI - kshlm & csim to set up faux/pseudo user email for gerrit, bugzilla, github 12:19:21 <hagarth_> almost there? 12:19:30 <kshlm> We were doing it right now. 12:19:51 <kshlm> csim found that we could just add ssh-keys to repositories directly. 12:20:02 <kshlm> We're testing if it works now. 12:20:15 <kshlm> This was for github only though. 12:20:26 <hagarth_> kshlm: cool, github is the most pressing right now 12:20:43 <kshlm> I think we'll have more updates next week. 12:21:05 <hagarth_> kshlm: ok, carrying forward this for tracking 12:21:12 <hagarth_> #action kshlm & csim to set up faux/pseudo user email for gerrit, bugzilla, github 12:21:30 <hagarth_> next AI - Need to decide if fixing BSD testing for release-3.6 is worth it. 12:21:38 <hagarth_> raghu: ? 12:21:43 <kshlm> This was on no one. 12:21:53 <hagarth_> what is this about? 12:22:06 <kshlm> regression fails for 3.6 on netbsd. 12:22:26 <hagarth_> I see, ok. what is Emmanuel's take on this? 12:22:36 <kshlm> This is mainly because the changes to regression framework haven't been backported. 12:22:58 <rastar> is this only about bad-tests list or something else? 12:23:12 <raghu> hagarth_: many netbsd fixes did not go to 3.6 12:23:20 <kshlm> I think emmanuel is okay if someone can backport the framework changes. 12:23:34 <raghu> so it was decided to consider only linux regressions runs for merging the patches 12:23:47 <hagarth_> raghu: ok 12:24:00 <hagarth_> kshlm: the framework code should be branch neutral right? 12:24:21 <kshlm> hagarth_, Should be. I've not checked though. 12:24:28 <rastar> hagarth_: it is not 12:24:36 <rastar> even that is version controlled 12:24:51 <kshlm> I think it'll mainly be the bad tests list that would differ. 12:25:06 <rastar> does it make sense to create a separate git for framework? 12:25:10 <hagarth_> rastar: right, I was wondering if we could copy files from framework/master to framework/release-3.6 12:25:37 <hagarth_> and if that would be the backport 12:25:44 <rastar> yes, but not all would apply 12:25:47 <kshlm> hagarth_, we'd need to test that out. 12:25:57 <rastar> new tests added for new features are not backported 12:26:24 <hagarth_> rastar: I am only thinking of .rc files and the likes, not .t units 12:27:01 <hagarth_> unless there are specific changes done in .t files for NetBSD, I think it would work 12:27:03 <rastar> hagarth_: ok, except bad_tests() function nothing else in .rc should refer to .t files 12:27:35 <rastar> would someone be willing to get the initial diff? I am out of office next week 12:27:41 <kshlm> So what do we do now? Do we have a volunteer to do this? 12:27:50 <hagarth_> rastar: I like the idea of keeping the basic framework branch neutral. let us see how we can get there. 12:28:23 <rastar> hagarth_: I will look into the branch neutral framework 12:28:31 <hagarth_> rastar: ok cool 12:29:14 <hagarth_> kshlm: let us track this.. I will attempt a brute force copy of rcs and send out a patch to see if it helps. 12:29:41 * msvbhat brb 12:29:42 <kshlm> Um, so the ai is both on you and rastar? 12:30:02 <hagarth_> kshlm: go ahead 12:30:20 <hagarth_> though we will be dealing with different aspects of the problem 12:30:33 <rastar> kshlm: yes, separate AIs would be better 12:31:05 <kshlm> #action hagarth to check if a bruteforce approach to backporting regression framework will work 12:31:24 <kshlm> #action rastar to go the more elegant way. 12:31:55 <hagarth_> next AI - rastar to close the glusterfs-3.7.6 tracker 12:32:35 <rastar> thats done 12:32:40 <hagarth_> rastar: cool, thanks. 12:33:00 <hagarth_> next AI - rastar Will call for volunteers for 3.7.7 release-manager 12:33:15 <rastar> I sent the mail but no replies yet 12:33:38 <rastar> ggarg was interested but he is not a maintainer yet 12:33:55 <rastar> any maintainer interested to work with ggarg on this? 12:34:19 <hagarth_> rastar: ok, if nobody does by end of this week .. I will have to spin my magic bottle & let it suggest a name ;). 12:34:35 <atinm> hagarth_, :) 12:34:35 <rastar> hagarth_: ;) 12:35:02 <hagarth_> ok, let us defer this. 12:35:20 <hagarth_> next AI - amye to get on top of disucssion on long-term releases. 12:35:36 <hagarth_> wonder what this is about? 12:35:51 <kshlm> I had a discussion about this with amye after the last meeting. 12:35:59 <kshlm> This will take time. 12:36:06 <kkeithley_> this is about a major community contributor who wants a gluster LTS release 12:36:21 <kkeithley_> not sure I can say their name. 12:36:58 <hagarth_> kshlm, kkeithley_: ok, should we track this AI? 12:37:00 <kkeithley_> we had a f2f meeting with them in Westford a couple weeks back. That's high on their list 12:37:22 <kshlm> I think we should. 12:38:20 <kkeithley_> yes, we should track it. I'm not sure how we decide to do it or not. Gluster Board? 12:38:27 <jdarcy> I'll take an AI to bug Amye about this. 12:38:28 <kkeithley_> Some other decision process? 12:38:45 <hagarth_> ok.. I think we need to at least have one meeting a month in a TZ that is amye friendly and have her host this. 12:39:09 <hagarth_> jdarcy: thanks! 12:39:26 <hagarth_> #action jdarcy to bug amye about disucssion on long-term releases 12:39:43 <hagarth_> next AI - Need to discuss how we do review-triages 12:39:52 <hagarth_> whose AI/topic is this? 12:40:13 <kkeithley_> mine, sort of 12:40:18 <kshlm> hagarth_, No one in particular. 12:40:32 <kshlm> But kkeithley_ did start the conversation which lead to it. 12:40:33 <kkeithley_> We have lots of reviews in gerrit that languish 12:41:03 <kkeithley_> some from community members who probably get frustrated by lack of action 12:41:19 <hagarth_> kkeithley_: I agree, this is a legitimate problem. 12:41:42 <kkeithley_> There was a rumor of sorts that Gartner is/was saying that the gluster community is stagnant, compared to ,e.g., Ceph 12:41:50 <kkeithley_> Maybe these are related 12:42:12 <hagarth_> kkeithley_: Gartner with their infinite wisdom can say a lot of things ;) 12:42:22 <hagarth_> however this is something that we need to address. 12:42:24 <kkeithley_> well, take it for what it's worth 12:43:00 <hagarth_> I have thought about this problem previously, some possible solutions: 12:43:01 <kkeithley_> It makes Red Hat Product Management sit up and take notice, and say things to us (me). 12:43:19 <hagarth_> 1. Enforce aging for patches in gerrit 12:43:35 <jdarcy> Just about the only time they give a $#@! about upstream. 12:44:16 <hagarth_> 2. All of us ensure that we review patches for which we get added as reviewers - this should encourage contributors to pull in the right reviewers for their patches. 12:44:25 <jdarcy> I think I suggested last time this came up that we arbitrarily designate a weekly top five (or top ten, whatever) patches for the "senior staff" to review. 12:45:12 <kkeithley_> We could have a triage-style review meeting, along the lines of the bug triage 12:45:12 <jdarcy> hagarth_: Regarding your point 2: for some of us that's going to be a *gigantic* stack of patches. 12:46:00 <hagarth_> jdarcy: agree. 12:46:13 <jdarcy> Even if I spent 100% of my work time reviewing patches to which I've been added, it would take me months to get through them. 12:46:30 <hagarth_> kkeithley_: should we club the bug + review triage meetings? 12:46:38 <rastar> Suggestion: Custom project dashboard which shows per component patches. I think gerrit supports it. 12:47:03 <hagarth_> rastar: that would be nice to have too. we'll get to know what areas need attention. 12:47:11 <hagarth_> rastar: do you have a sample of that? 12:47:29 <rastar> here is the doc: http://siekiera.mimuw.edu.pl:8082/Documentation/user-dashboards.html#project-default-dashboard 12:47:35 <rastar> haven't tried it yet 12:47:40 <hagarth_> rastar: cool, will check that out. 12:47:43 <kkeithley_> "clubbing" works as long as we can get through the bugs with time left to do reviews. 12:47:57 <kshlm> rastar, Is that even supported with our gerrit version? 12:48:28 <rastar> kshlm: not sure, but then I would favour upgrading gerrit over dropping idea 12:48:32 <kshlm> rastar, nvm. It is. 12:48:58 <rastar> i don't really like 2.9.x version 12:49:16 <hagarth_> should we start a discussion on -devel and/or -maintainers about review backlog clearance? 12:49:26 <kshlm> hagarth_, Yes. 12:49:37 <kshlm> That was the basic goal of the AI. 12:49:53 <hagarth_> ok, I will start that thread. 12:49:55 <kkeithley_> yes 12:50:00 <Humble> http://review.gluster.org/#/q/message:glusterd+AND+status:open -> something like this can filter gerrit with a tag and patch status 12:50:07 <kshlm> Thanks hagarth_ 12:50:27 <hagarth_> #action hagarth_ to start a thread on review backlog 12:50:47 <hagarth_> ok, we are done reviewing & discussing our long list of AIs :). 12:50:53 <hagarth_> moving on to next topic 12:51:02 <atinm> we are running out of time :( 12:51:05 <hagarth_> #topic GlusterFS 3.7 12:51:17 <kshlm> 50 minutes for AIs! 12:51:20 <hagarth_> atinm: yes, noted that. will try to skim over the usual topics. 12:51:41 <hagarth_> 3.7.7 needs a maintainer and will be released in the next 2-3 weeks. 12:51:55 <hagarth_> I will spin my magic bottle if we don't find a maintainer by end of this week. 12:52:04 <hagarth_> anything else on 3.7? 12:52:19 <hagarth_> guess not, moving on 12:52:24 <hagarth_> next topic - GlusterFS 3.6 12:52:29 <raghu> I have made 3.6.7 12:52:46 <hagarth_> raghu: thanks for the update, I see the tag. packages are to be built? 12:52:58 <raghu> yeah 12:53:14 <hagarth_> raghu: cool, thank you! 12:53:16 <kkeithley_> tag. packages are to be built? meaning? 12:53:21 <kshlm> 3.6.7 was released 3 times. 12:53:26 <hagarth_> kshlm: ? 12:53:32 <kshlm> Just an observation. 12:53:57 <kkeithley_> oh, sorry. too early for me. Yes, I'll be calling for volunteers to build packages 12:54:01 <raghu> sorry about that. I had added my email as well. Since I did not get the mail, I thought, the build might have failed and again retriggered it twice 12:54:01 <kshlm> raghu, triggered the release job 3 times. 12:54:08 <hagarth_> kkeithley_: I see the tag in git. missed being clear there. 12:54:20 <hagarth_> raghu: ok 12:54:23 <hagarth_> kkeithley_: thanks 12:54:43 <hagarth_> so announcement of 3.6.7 will happen next week. 12:54:46 <hagarth_> moving on 12:54:57 <hagarth_> skipping 3.5 as ndevos is not around today. 12:55:15 <hagarth_> #topic GlusterFS 3.8 12:56:07 <hagarth_> we need to nail down a schedule for 3.8. I will be starting a discussion around this and possibly host a meeting next week with owners of proposed features to get a better picture. 12:56:32 <hagarth_> #action hagarth_ to start a discussion around 3.8 release dates. 12:56:42 <hagarth_> any questions on 3.8? 12:57:06 <kshlm> None yet. 12:57:13 <hagarth_> guess not, moving on. 12:57:20 <hagarth_> next topic - GlusterFS 4.0 12:57:52 <hagarth_> Jeff, Atin - any updates here? 12:58:02 <atinm> hagarth_, work continues :) 12:58:09 <jdarcy> Not much, other than what I'll send later today. 12:58:33 <atinm> hagarth_, GlusterD 2.0 is focusing on addressing API comments and coming up with the design doc 12:58:38 <jdarcy> Still trying to put together a meeting with our Large External Contributor about some issues they've raised. 12:58:46 <hagarth_> jdarcy, atinm: ok 12:58:59 <atinm> hagarth_, I got to know that DHT2 is focusing on mkdir to work 12:59:20 <hagarth_> atinm: ok cool 12:59:23 <atinm> hagarth_, and Avra has sent couple of patches on code generation (server & client) 12:59:35 <hagarth_> nice, we are making good progress! 12:59:56 <hagarth_> anything else here? 12:59:59 <atinm> hagarth_, and most importantly experimental directory is now in the repo, all 4.0 patches should land up there 13:00:19 <hagarth_> atinm: great! 13:00:47 <hagarth_> if no questions here, moving on to next topic 13:00:50 <hagarth_> #topic Open Floor 13:01:20 <hagarth_> I see two topics in etherpad 13:01:27 <hagarth_> apart from the weekly reminder ones 13:01:30 <hagarth_> 1. We need to make it a requirement that all new code get clean Coverity, cppcheck, clang analyse, and valgrind runs before merging. 13:01:31 <kshlm> Those were carried forward from last week. 13:01:49 <kkeithley_> very briefly... I want to draw a line in the sand that new features should be cppcheck, coverity, valgrind "clean" 13:02:08 <hagarth_> kkeithley_: +1 13:02:26 <kkeithley_> clang analyze for bonus points! 13:02:28 <hagarth_> kkeithley_: can you start a ML discussion on this? 13:02:42 <kkeithley_> sure 13:03:04 <hagarth_> kkeithley_: once we bump up gerrit, I would like to integrate it with new avatar of sonar (forget what it is called now) 13:03:25 <kshlm> sonraqube 13:03:26 <kkeithley_> and we have resources. If people need help using them they should ask! 13:03:41 <hagarth_> #action kkeithley_ to start a ML discussion on code hygiene for new features. 13:03:49 <hagarth_> kshlm: thanks! 13:03:54 <hagarth_> topic #2. Find an accountable way to ask for backports 13:04:04 <hagarth_> whose topic is this? 13:04:09 <kkeithley_> we have nightly cppcheck, clang analyze, and coverity builds. I bet nobody ever looks at them. 13:04:16 <kshlm> This came up in the 3.5 discussion last week. 13:04:37 <hagarth_> kkeithley_: +1 13:04:43 <jdarcy> So how do we enforce this "hygiene" requirement? Require developers to run all these things manually? Provide some automation? Have certain people run the tools separately and convey the result as -2 review comments? 13:05:18 <kshlm> We had no changes lined up for 3.5, so kkeithley_ suggested it could be because we don't have an accountable way to ask for backports. 13:05:47 <hagarth_> jdarcy: sonarqube can automatically flag static analysis warnings for diffs/patches in gerrit. I would like to see most of it being automated. 13:06:17 <jdarcy> OK, that's one. How about the other five Kaleb mentioned? 13:06:36 <kkeithley_> we tried having people edit the wiki. Not everyone can edit bugs, so adding to the tracker bug is out. Maybe just a backport etherpad on public.pad.fsfe.org 13:07:03 <hagarth_> kkeithley_: sounds like a good idea to me. 13:07:27 <hagarth_> jdarcy: do some post analysis on the runs that kkeithley_ does with those tools? 13:08:15 <hagarth_> unless it can be automated, it is going to be hard to enforce the "hygiene" requirement IMO. 13:08:38 <kkeithley_> wrt to "hygiene" at the most basic level I'd say developers should run the tools themselves, look at the results, and self declare that their patch is clean in the review comments. 13:09:02 <jdarcy> hagarth_: Perhaps. I think we'll need some automation to keep this from being a big time-suck either for every developer or for those few running the checks. 13:09:42 <hagarth_> kkeithley_, jdarcy: let us carry forward this discussion in the mailing list thread that kkeithley_ will start. 13:10:04 <jdarcy> Also some documentation/education might be nice. Regardless of whether we think they should, a lot of our developers don't already know how to run these tools or interpret their results. 13:10:06 <kkeithley_> because, e.g. coverity, reports hundreds of issues all over the tree. It only takes about five minutes for a coverity run, then check that there are no (new) issues in the code you're submitting 13:10:13 <hagarth_> jdarcy: +1 13:10:21 <kkeithley_> agreed, re: mailing list 13:10:51 <kkeithley_> re: mailing list discussion 13:11:24 <hagarth_> anything else for today? 13:11:48 <justinclift> Meh 13:12:14 <hagarth_> guess not, thank you all for attending today. Happy Thanksgiving to everyone and see you on the other side of Thanksgiving! 13:12:17 <kshlm> hagarth_, just the weekly reminders. 13:12:26 <justinclift> ;) 13:12:35 <hagarth_> Weekly reminder to announce Gluster attendance of events: https://public.pad.fsfe.org/p/gluster-events 13:12:43 <hagarth_> REMINDER to put (even minor) interesting topics on https://public.pad.fsfe.org/p/gluster-weekly-news 13:12:51 <hagarth_> kshlm: done :) 13:12:54 <kshlm> hagarth_, :) 13:12:57 <hagarth_> #endmeeting