12:01:24 <JustinClift> #startmeeting Weekly Gluster Community IRC Meeting
12:01:24 <zodbot> Meeting started Wed Apr 22 12:01:24 2015 UTC.  The chair is JustinClift. Information about MeetBot at http://wiki.debian.org/MeetBot.
12:01:24 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
12:01:37 <JustinClift> k... who do we have? :)
12:01:46 <JustinClift> Btw - Etherpad: https://public.pad.fsfe.org/p/gluster-community-meetings
12:01:50 * tigert is in
12:01:55 * jimjag is here
12:02:01 * kkeithley is here
12:02:06 * jdarcy is lurking somewhere
12:02:09 <JustinClift> If you're on the Etherpad, please put your name in the top right box thing so we know who's typing what :)
12:02:22 * hagarth1 is in a different meeting, partially here too
12:02:25 * overclk is here
12:03:26 <JustinClift> Cool.
12:03:31 * spot is here
12:03:32 <JustinClift> Agenda items from last week:
12:03:40 <JustinClift> #topic Agenda items from last week
12:03:51 <JustinClift> "jdarcy should probably still write the segfault identifier"
12:03:58 <JustinClift> Is this still needed?
12:04:13 <jdarcy> Still not done.  Might be less important with the logging-crash fix.
12:04:24 <JustinClift> k.  Drop the AI then?
12:04:37 <jdarcy> Sure.  We can always reinstate if necessary.
12:04:39 * kshlm /o
12:05:03 * raghu is here
12:05:09 <JustinClift> "overclk will get details of the overlay xlator from hagarth"
12:05:19 <JustinClift> overclk: Complete?
12:05:39 <overclk> hagarth has a feature page for that
12:05:43 <overclk> https://www.gluster.org/community/documentation/index.php/Features/Overlay
12:05:51 <JustinClift> Heh, I wondered if that would count ;)
12:06:05 <raghu> overclk: I am also interested in overlay
12:06:11 <JustinClift> overclk: Have you reviewed the feature page?
12:06:13 <overclk> raghu, cool!
12:06:17 * soumya_ is here
12:06:18 <overclk> JustinClift, nope, not yet.
12:06:30 <overclk> JustinClift, I'll probably do that this week.
12:06:36 <JustinClift> #action overclk will review the Overlay feature page by next meeting :)
12:06:38 <kshlm> overlay xlator looks interesting.
12:06:53 <JustinClift> "jdarcy to backport http://review.gluster.org/#/c/10157/ to release-3.7"
12:06:58 <JustinClift> That's done:
12:07:03 <JustinClift> http://review.gluster.org/#/c/10255/
12:07:09 <JustinClift> It needs reviewers though.
12:07:12 <JustinClift> Any volunteers?
12:07:18 <JustinClift> It's pretty important
12:07:48 <jdarcy> Closely related: should probably backport http://review.gluster.org/#/c/10322/ as well
12:08:29 <JustinClift> #action jdarcy to backport http://review.gluster.org/#/c/10322/ to release-3.7 branch
12:09:21 <JustinClift> jdarcy: What's a good way to get more reviewer volunteers for you?
12:09:31 <jdarcy> AFAICT that would increase our regression-test success rate from ~5% to ~80%
12:09:42 * ndevos arrives late...
12:09:46 <jdarcy> JustinClift: I'll go into Nag Mode.
12:09:55 <JustinClift> k.  pls do
12:10:11 <JustinClift> #action jdarcy to transform into an old horse for a week
12:10:14 <jdarcy> If it's still unresolved when I'm BLR physically, I'll knock heads.
12:10:16 <atinmu> I am seeing the auto retry of failed test cases in netbsd is not helping me
12:10:17 <JustinClift> ^ that's what you meant yeah? :)
12:10:32 <atinmu> Every time I trigger the same test fails in 2nd attempt :(
12:10:39 <jdarcy> atinmu: Yeah, it seems like the retry always fails exactly the same way as the original.
12:10:51 <atinmu> Am I too unlucky or we have a bug here ??
12:11:28 <JustinClift> Could be either :)
12:11:37 <atinmu> JustinClift, :)
12:11:43 <JustinClift> atinmu: Would you be ok to bring that up on gluster-devel mailing list?
12:11:59 <jdarcy> I think a lot of failures are timing-related, and on a given slave over a short interval the timing doesn't change much.
12:12:14 <atinmu> JustinClift, I will do that
12:12:49 <JustinClift> #action atinmu to raise the 2nd failures on NetBSD not helping much, on gluster-devel mailing list
12:12:53 <JustinClift> Cool :)
12:13:47 <JustinClift> jdarcy: Is there some way for us to be able to measure the timing-ness of a VM so we can get a better idea of what causes what?
12:13:55 <JustinClift> jdarcy: "timing-ness" being a crap word
12:14:01 * JustinClift doesn't know a better one
12:14:34 <JustinClift> eg if a VM is going a certain amount of slow due to it's host being under load or something... can we assign that a measure?
12:14:56 <jdarcy> Maybe, but it doesn't lead to a solution.
12:15:02 <JustinClift> So we can tell if certain tests fail if the vm is in slowness of  a particular range
12:15:11 <JustinClift> k, moving on :)
12:15:12 <jdarcy> What we sort of need to do is retry on a different machine, but that'll be complicated.
12:15:44 <JustinClift> "hagarth to help raghu with merging fixes for build failures"
12:15:53 <JustinClift> It's marked as done in the Etherpad
12:15:59 <JustinClift> "jdarcy to hold meeting on 17 April"
12:16:13 <JustinClift> This happened, and a good time was had by all
12:16:17 <jdarcy> Done, minutes here.  http://meetbot.fedoraproject.org/gluster-meeting/2015-04-17/gluster-meeting.2015-04-17-12.05.html
12:16:27 <JustinClift> Thanks jdarcy :)
12:16:28 <atinmu> JustinClift, mail shooted :)
12:16:34 <JustinClift> "spot to announce summit"
12:16:39 <JustinClift> atinmu: Awesome :)
12:17:08 <spot> the fact that we have 39 attendees confirmed means this is effectively announced, but we'll include it in our PR at the end of the month.
12:17:15 <JustinClift> spot: Ironically I don't remember seeing that
12:17:18 <JustinClift> Ahhh yeah
12:17:31 <jdarcy> My family's getting tired of me singing Bar-ce-lo-na like the opening phrase of Beethoven's Fifth.
12:17:43 * spot is hesitant to scream too loud about it when we really can't take any more people. :)
12:17:48 <tigert> :)
12:18:02 <krishnan_p> jdarcy, would like to hear you do that in Barcelona ;-)
12:18:05 <spot> i think this action item can be marked as complete unless anyone disagrees.
12:18:06 * msvbhat arrives late
12:18:15 <jdarcy> krishnan_p: After a sangria or two, maybe.
12:18:41 <krishnan_p> jdarcy, I was going to suggest before every feature that you 'announce'.
12:18:47 <JustinClift> Heh
12:19:00 <JustinClift> "JustinClift to get basic forge v2 operational by 15 April"
12:19:05 <JustinClift> Well, that didn't happen
12:19:20 <JustinClift> Basic stats collection code was written
12:19:28 <JustinClift> But I need to get projects migrated over
12:19:38 <JustinClift> So... it'll take a bit more time.
12:19:53 <JustinClift> #action JustinClift to get basic forge v2 operational by next meeting
12:20:03 <JustinClift> #topic 3.7
12:20:12 <JustinClift> How are we looking with this?
12:20:13 <kshlm> You did upgrade and fix gerrit! That's an awesome excuse though.
12:20:27 <JustinClift> kshlm: it's still kinda broken tho :(
12:20:47 <kshlm> Works for me.
12:20:47 * JustinClift is still doing lots of manual work in the background and will prob upgrade it again this wkend
12:20:51 <JustinClift> ;)
12:21:13 <JustinClift> hagarth1: 3.7 status update?
12:21:22 <jdarcy> Everybody go tell Justin's boss how awesome he is.
12:21:37 <JustinClift> For breaking Gerrit?  Erm... maybe not
12:21:58 <jdarcy> You can't make an omelet...
12:22:00 <JustinClift> ;)
12:22:09 <hagarth1> JustinClift: propose to push 3.7.0 by a week
12:22:19 <hagarth1> sent a note on the mailing lists a few hours back
12:22:19 <JustinClift> Sounds good to me
12:22:21 <jdarcy> Only one?
12:22:35 <JustinClift> It was scheduled for 1 week from now... and we haven't had any test days
12:22:40 <hagarth1> jdarcy: yes!
12:22:59 <hagarth1> JustinClift: the previous test days did not yield anything of much significance, I wonder if we should have one.
12:23:15 <JustinClift> We've not pushed out any alpha/beta have we?
12:23:17 <jdarcy> I'm just thinking, with the number of features still in flux and the Jenkins backlog, do we want to consider two weeks instead.
12:23:33 * JustinClift mutters ... month ... under his breath
12:23:50 <hagarth1> I am looking at a release early, release often strategy for 3.7
12:23:53 <dlambrig_> …certain critical QE members getting married and away…
12:24:10 <hagarth1> so I am fine with making as many releases in 3.7.x to stabilize
12:24:23 <JustinClift> hagarth1: Cool.  Lets get some alpha/beta releases out the door then :)
12:24:30 <hagarth1> JustinClift: will do a beta this week
12:24:34 <JustinClift> :)
12:24:47 <JustinClift> #action hagarth to release 3.7.0beta1 this week
12:24:52 <kkeithley> what's up with alpha0?
12:25:00 <kkeithley> tagged, but no release?
12:25:02 <hagarth1> I will drop a note later on what all we need for 3.7.0qa later today.
12:25:04 <kshlm> I think we're still too broken to do a beta
12:25:08 <hagarth1> kkeithley: the nightly rpms have that
12:25:16 <hagarth1> kshlm: it is beta1, not the final beta
12:25:16 <krishnan_p> hagarth1, if we release early, we need to add a note in our release notes regarding our confidence of things that have gone in too.
12:25:18 <JustinClift> Well, we have an alpha1 dir on the website with no content: http://download.gluster.org/pub/gluster/glusterfs/qa-releases/3.7.0alpha1/
12:25:21 <hagarth1> krishnan_p: +1
12:25:24 <ndevos> hagarth1: please no 3.7.0qa, keep alpha/beta/... :)
12:25:31 <hagarth1> ndevos: certainly
12:25:52 <JustinClift> 3.7.0 - "Run this if you're feeling lucky" release ?
12:26:14 <kkeithley> um, okay. Nightly builds aren't in Fedora.  Or maybe we don't care about that?
12:26:24 <hagarth> JustinClift: which bleeding edge releases are not in that manner?
12:26:29 <JustinClift> :)
12:26:30 <kkeithley> Or on download.gluster.org
12:26:42 <ndevos> kkeithley: there should be nightly builds for fedora?
12:26:59 <ndevos> http://download.gluster.org/pub/gluster/glusterfs/nightly/glusterfs-3.7/ is non-empty
12:27:00 * JustinClift is a bit ambivalent about Fedora.  If other ppl have opinions tho... ?
12:27:02 <hagarth> our community has a persona where users pick up only released versions
12:27:11 <hagarth> and for the right obvious reasons
12:27:12 <kkeithley> no, alpha0 should maybe be in fedora rawhide?
12:27:25 <ndevos> oh, no!
12:27:28 <hagarth> so it doesn't make sense to release a lot of alpha/beta releases unless we get *real* feedback.
12:27:55 <ndevos> kkeithley: lets try to stick with GA releases, maybe beta
12:28:08 <JustinClift> hagarth: Would that mean our community members really, really don't want unstable release then?
12:28:33 <hagarth> JustinClift: unless we get feedback from the community, we know for what usecases we would be unstable.
12:28:42 <hagarth> s/know/would not know/
12:28:43 <kkeithley> well, this is not the case with f23 yet, but e.g. to "reserve" ganesha-2.2.0 in f22 I had been putting their RCs into Fedora.
12:28:47 <jdarcy> It's a chicken and egg problem.
12:28:52 <JustinClift> Yeah
12:28:57 <ndevos> JustinClift: storage is often a critical system, users tend to be conservative on updating :)
12:29:01 <hagarth> yes
12:29:18 <hagarth> I run tests that I can but cannot cover all possible workload simulations
12:29:21 <JustinClift> Yeah, I tend to fall in that category with my storage expectations too
12:29:36 <JustinClift> k... any AI's from this?
12:30:06 <ndevos> kkeithley: ganesha was already in the release-candidate state, right?
12:30:44 <ndevos> #argeed releasing 3.7.0 will be delayed for one week
12:31:02 <ndevos> uh, lets try that again
12:31:06 <ndevos> #agreed releasing 3.7.0 will be delayed for one week
12:31:07 <kkeithley> Historically we've been putting alpha and beta releases on http://download.gluster.org/pub/gluster/glusterfs/qa-releases/    and the Fedora and RHEL RPMs are built in Fedora Koji.
12:31:41 <JustinClift> Moving on...
12:31:53 <ndevos> kkeithley: right, but that is not *in* Fedora, those are only scratch builds and do not need updates in fedpkg/dist-git
12:32:15 <JustinClift> Should we move on or ?
12:32:20 <kkeithley> right, but we need a released tarball to build with!
12:32:29 <JustinClift> You guys look like you're still going, and it sounds important :)
12:32:51 <JustinClift> Discuss here or mailing list?
12:33:01 <ndevos> kkeithley: yes, and we get one when hagarth sais so :) alpha0 is only a marker for the branch point of release-3.7
12:33:30 * ndevos thinks its fine to move to the next topic
12:33:39 <JustinClift> #topic 3.6
12:33:57 <raghu> I have made 3.6.3 and the tarball can be found here (http://bits.gluster.org/pub/gluster/glusterfs/src/glusterfs-3.6.3.tar.gz).
12:34:20 <JustinClift> Cool.  When are packages and announcement expected? :)
12:34:22 <kshlm> I didn't see an announcement!
12:34:26 <raghu> I have updated humble, niels, lala and kaleb. Will send a mail once the rpms are ready
12:34:46 <kshlm> Should the release maintainer wait for the packages to be ready before making an announcement?
12:34:48 <JustinClift> raghu: Likely today or by end of day tomorrow or similar?
12:34:53 <JustinClift> kshlm: Yes
12:35:06 <raghu> JustinClift: yeah.
12:35:07 <kshlm> Hmm...
12:35:12 <JustinClift> kshlm: Otherwise ppl get excited and then find there's nothing to download when they want to
12:35:24 <ndevos> humble and lala normally build those rpms... lets wait until they made them available
12:35:29 <kshlm> I thought package maintainers would listen for the release annoncement to make packages
12:35:52 <JustinClift> We seem to let them know directly, before that
12:36:11 <JustinClift> Which lets us do announcements with things that ppl can actually use when they read about it
12:36:13 <ndevos> kshlm: no, the maintainers send a heads-up to the package maintainers, who then start building for a more  coordinated announcement
12:36:16 <JustinClift> (which is better marketing)
12:36:40 <JustinClift> raghu: Sounds good
12:36:46 <ndevos> this prevents the "where are the packages?" questions :)
12:36:50 <JustinClift> raghu: What's the plan for 3.6.4 ? :)
12:36:53 <kshlm> So do we just have RPMs ready, or do we have debs as well?
12:36:53 <kkeithley> package maintainers have heard from raghu. We're waiting for them to make the Fedora and RHEL RPMs. Debian and other packages are also in the works for 3.6.3
12:37:27 <kshlm> Cool. I didn't know how the announcements work. I was just curious.
12:37:50 <JustinClift> kshlm: All good :)
12:37:51 <raghu> JustinClift: I am going create a new tracker bug for 3.6.4 today and there are some patches that have been sent for release-3.6. I think they r suffecient to make a beta1
12:38:06 <JustinClift> Cool.  ETA for 3.6.4beta1?
12:38:32 <raghu> I am planning to do it by the end of next week.
12:38:54 <JustinClift> Excellent :)
12:39:05 <JustinClift> #action raghu to create new 3.6.4 tracker bug today
12:39:20 <JustinClift> #action raghu to release 3.6.4beta1 by the end of next week
12:39:29 <JustinClift> raghu: Anything else for 3.6 topic?
12:39:43 <raghu> JustinClift: nope.
12:40:34 <JustinClift> k.  It sounds like you've been putting in a lot of work for 3.6.x stuff.  Good stuff. :)
12:40:47 <JustinClift> #topic 3.5
12:41:14 <ndevos> packages for 3.5.4beta1 are almost ready, I think
12:41:26 <ndevos> they need some testing, and I hope to do that later today
12:41:56 <ndevos> when that passes, we can get more  testers - hopefully
12:42:03 <raghu> He he. There were some last minute bugs that delayed the release
12:42:35 <JustinClift> Should we expect a release announcement today/tomorrow for them?
12:42:43 <ndevos> I already have one patch that needs to get included in 3.5, so I think a beta2 will happen next week
12:43:14 <JustinClift> k.  So, do the release announcement for the one next week then?
12:43:27 <ndevos> 3.5.4beta1 should get announced when the packages are ready, hopefully later this week
12:43:32 <JustinClift> k
12:43:49 <JustinClift> #action ndevos to do release announcement for 3.5.4beta1 when the packages are ready
12:44:00 <JustinClift> Hopefully this week, but guess we'll see ;)
12:44:02 <ndevos> thats all :)
12:44:06 <JustinClift> :)
12:44:13 <JustinClift> #topic 4.0
12:44:24 <JustinClift> jdarcy: This be you I think
12:44:46 <jdarcy> Nothing we didn't cover earlier.
12:44:50 <JustinClift> k
12:44:53 <JustinClift> Moving on...
12:45:03 <JustinClift> #topic Other Agenda Stuffs
12:45:17 <JustinClift> "Consider alternating time zones"
12:45:42 <JustinClift> Someone wrote: "There are members of the other hemisphere that would like to be active in the community but cannot attend meetings at this hour. I propose alternating meetings by 12 hours, 0:00 and 12:00 UTC."
12:45:51 <JustinClift> overclk: Your suggestion? ^
12:46:16 <JustinClift> The concept sounds ok
12:46:25 <JustinClift> eg having two meetings that alternate
12:46:42 <jdarcy> I'm OK with it.
12:46:51 <JustinClift> 12 hours may not be the right gap, but we can fine tune it
12:46:54 * ndevos will alternate with the meetings in that case :)
12:47:03 * JustinClift too probably
12:47:08 <JustinClift> But, we'll see
12:47:21 <JustinClift> I guess we can try it out, and see if it's a win or not
12:47:26 <JustinClift> Anyone think it's a bad idea?
12:47:48 <ndevos> well, it all depends on who will attend the "other" meeting
12:47:59 <JustinClift> Yep
12:48:14 <JustinClift> Hands up for ppl that would attend a meeting 12 hours earlier... ?
12:48:18 <ndevos> do we know who (or how many) would attend that can not attend now?
12:48:33 <ndevos> lol, those people are probably not there now, JustinClift :D
12:48:41 <JustinClift> Maybe we should raise this on gluster-devel then
12:48:50 <ndevos> yes, and gluster-users
12:48:54 <JustinClift> And yeah, I realised that after I typed it
12:49:00 <JustinClift> "sounded better in my head" :D
12:49:07 <ndevos> and the board list, maybe they are in the wrong timezone too?
12:49:08 * spot is normally putting children to bed at that time, but I can make it work.
12:49:38 <JustinClift> k.  I'll email out the suggestion, as the author seems to be keeping anon :/
12:50:10 <JustinClift> #action JustinClift to email gluster-users/devel/board with proposal for alternating meeting times
12:50:32 <JustinClift> Anyone else have topic/item they want to bring up? :)
12:50:37 <jdarcy> So who's going to be in BLR (who isn't usually) next week?
12:50:50 * ndevos just came back...
12:51:38 <JustinClift> jdarcy: Guess it'll be just you
12:51:50 <JustinClift> k, this meeting's a wrap
12:51:57 <JustinClift> Thanks everyone :)
12:52:06 <kkeithley> Ric is en route to BLR, maybe he's already there
12:52:07 <JustinClift> #endmeeting