15:01:23 <hagarth> #startmeeting
15:01:23 <zodbot> Meeting started Wed Dec 11 15:01:23 2013 UTC.  The chair is hagarth. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:01:23 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
15:01:23 <glustermeetbot> Meeting started Wed Dec 11 15:08:50 2013 UTC.  The chair is hagarth. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:01:23 <glustermeetbot> Useful Commands: #action #agreed #help #info #idea #link #topic.
15:01:31 <hagarth> hi all
15:01:48 <hagarth> #topic Rollcall & AI follow up
15:01:57 <hagarth> who do we have here now?
15:02:06 * ndevos _o/
15:02:06 <johnmark> abyss: this is a channel for meetings. but at the end, you can direct a question to someone and they can hopefully help you
15:02:29 <johnmark> abyss: if you can't find anyone, send me a private note at johnmark (at) gluster dot org
15:02:43 <johnmark> ndevos: nice
15:02:48 <hagarth> ok, moving on to action items from last week
15:02:58 <johnmark> Technicool: howdy!
15:03:04 <Technicool> morning
15:03:06 <hagarth> I have not been able to complete one AI
15:03:21 <Technicool> avati, the meeting starts at 7am SHARP!
15:03:22 <hagarth> i.e. follow up on gluster-devel about tests for libgfapi
15:03:23 <Technicool> ;)
15:03:33 <hagarth> intend doing that this week
15:03:39 <avati> Technicool :(
15:03:43 <johnmark> hagarth: awesome :)
15:03:47 <johnmark> avati: mornin'
15:03:48 <avati> Technicool: you too are in PST right?
15:03:55 <Technicool> avati, i joined seconds before you  ')
15:03:55 <hagarth> I will also have the documentation page setup by tomorrow
15:04:00 <johnmark> hagarth: thanks
15:04:00 <Technicool> avati, yes
15:04:11 * johnmark looks at AIs
15:04:11 <avati> Technicool: shame on you, for being earlier :p
15:04:23 <johnmark> http://titanpad.com/gluster-community-meetings
15:04:38 <hagarth> meeting minutes from last week - http://meetbot.fedoraproject.org/gluster-meeting/2013-12-04/gluster-meeting.2013-12-04-15.01.html
15:05:05 <hagarth> Technicool, johnmark: you seem to have one between the two of you :)
15:06:07 <hagarth> Technicool, johnmark: I think we can defer that AI to this week - doesn't impact the documentation hackathon this week
15:06:08 <johnmark> hagarth: I know I did at least one of them - just trying to retrace my steps :)
15:06:15 <johnmark> hagarth: ok
15:06:26 <hagarth> ok, moving on
15:06:29 <hagarth> #topic 3.5
15:06:37 <abyss> johnmark: OK. Thank you, I'll send you e-mail. OK? Because I don't know whos would help me here;) And I have gluster on production environments so it's quiet importat for me (and my company;))
15:07:02 <hagarth> The first test weekend did not produce a lot of bugs
15:07:13 <johnmark> abyss: ok. you're welcome
15:07:19 <abyss> thank you very much
15:07:20 <hagarth> that makes me feel that we have not got enough testing coverage for everything in 3.5.
15:07:48 <johnmark> hagarth: ok
15:07:55 <hagarth> suggestions on getting more testing coverage in the 2nd test weekend?
15:07:57 <johnmark> hagarth: then let's start ramping up for hte next one
15:08:10 <johnmark> hagarth: Ben England submitted some code for doing scale-out testing
15:08:11 * johnmark looks
15:08:27 <johnmark> hagarth: could be useful to more easily test in larger environments
15:08:32 <johnmark> hang on...
15:08:37 <hagarth> johnmark: ok
15:08:58 <hagarth> we have also slipped the 3.5 beta by two days now.
15:09:02 <ndevos> maybe provide a vm image that is pre-installed so that users can easier start testing?
15:09:02 <johnmark> hagarth: https://github.com/bengland2/smallfile
15:09:11 <johnmark> #link https://github.com/bengland2/smallfile
15:09:21 <hagarth> ndevos: that sounds like a good idea
15:09:28 <johnmark> ndevos: +1
15:09:46 <johnmark> hagarth: if hte DHT patches have been merged, I'm inclined to say unleash the krakken... er, beta
15:09:46 <hagarth> ndevos: can we both take that up as an action item for test day 2?
15:10:10 <Technicool> ndevos, i was thinking the same thing but maintaining the image would be cumbersome between all the releases
15:10:14 <ndevos> hagarth: yeah, I guess so
15:10:18 <johnmark> Technicool: true
15:10:22 <hagarth> #action ndevos and hagarth to look into a pre-packaged VM for test day 2
15:10:42 <Technicool> maybe the same problem but an automated test for docker could work as well
15:10:46 <johnmark> let's be honest, most people don't really pay attendtion to these things until the beta comes out
15:10:46 <hagarth> Technicool: I think the scope of such an image would be limited, get testing feedback and move on
15:10:51 <johnmark> Technicool: +1
15:10:56 <johnmark> hagarth: true
15:11:11 <hagarth> johnmark: right. we have all the patches in dht for 3.5.
15:11:13 <ndevos> Technicool: it would need to be an image for a specific test-day
15:11:30 <ndevos> what distribution(s) should such an image be?
15:11:44 <johnmark> ndevos: all possible :)
15:11:45 <hagarth> I am awaiting a  few geo-replication patches, one feedback that we got from test day 1 was that geo-replication is not working well.
15:11:53 <johnmark> hagarth: ok
15:11:58 <Technicool> fedora/rhel/debian/ubuntu i would guess
15:12:00 <johnmark> hagarth: who's working on them?
15:12:06 <johnmark> Technicool: +centos
15:12:10 <johnmark> oh rhel, haha
15:12:14 <johnmark> nm
15:12:15 <Technicool> ; )
15:12:16 <hagarth> ndevos: Ajeet Jha has sent out a few patches
15:12:21 <ndevos> -rhel, we cant distribute those bits :)
15:12:30 <hagarth> ndevos: they need to be reviewed and merged
15:12:41 <Technicool> ndevos rhel was a synonym for centos
15:12:47 <Technicool> and scientific linux
15:12:51 <johnmark> hagarth: ok. can we get the expected patches merged in by tomorrow or Friday?
15:12:56 <hagarth> avati: can you work on getting the geo-replication patches in?
15:12:57 <ndevos> Technicool: sure :)
15:13:06 <johnmark> Technicool: heh
15:13:15 <hagarth> johnmark: yes, we can do a beta as soon as the geo-replication patches are merged.
15:13:24 <avati> hagarth: yes i have started reviewing it yday
15:13:25 <ndevos> hagarth: not sure what I should do with those patches...
15:13:26 <johnmark> Technicool: should have been <air quotes>RHEL</air quotes>
15:13:31 <kkeithley> say what? we can't distribute community el6 rpms?
15:13:32 <Technicool> lol
15:13:34 <johnmark> avati: thanks
15:13:37 <johnmark> hagarth: cool
15:13:48 <johnmark> kkeithley: of course we can
15:13:49 <hagarth> #action avati to review and merge geo-replication patches
15:13:54 <ndevos> kkeithley: we cant distribue a rhel vm for testing (I guess)
15:14:06 <ndevos> *distribute
15:14:10 <johnmark> ndevos: right
15:14:14 <kkeithley> oh, sorry, I missed the rhel vm part
15:14:27 <hagarth> ok folks, here's the question on beta:
15:14:40 <hagarth> do we all think that we are ready to hit beta once geo-rep patches are in?
15:15:06 <johnmark> hagarth: I say yes
15:15:18 <ndevos> yes, I think so too
15:15:20 <johnmark> hagarth: because we won't get nearly enough testing until we say "it's beta"
15:15:26 <hagarth> ok, cool.
15:15:43 <hagarth> #info 3.5 beta to be out after geo-replication patches are merged
15:15:57 <hagarth> are there other patches that folks want to get into 3.5?
15:16:31 <hagarth> guess not
15:16:37 <ndevos> I would like to see the systemd pieces included... but I have not filed reviews for that yet
15:16:45 <kkeithley> my fedora spec merge/sync?
15:16:46 <ndevos> and neither did kkeithley, I think
15:17:00 <hagarth> kkeithley: I think we need to merge the fedora spec sync
15:17:12 <ndevos> no, the .service files that are in the fedora repo, move them to upstream
15:17:44 <kkeithley> you mean the glusterfsd.service (and glusterfsd init.d) files?
15:17:58 <hagarth> ndevos: what is the systemd change that you would like to see in 3.5?
15:18:20 <ndevos> yes, glusterd.service and glusterfsd.service - ultimately the init.d scripts too
15:18:51 <johnmark> hagarth: are there any deb packaging things that need to be merged?
15:18:52 <kkeithley> glusterd.service is already in
15:18:54 <johnmark> semiosis: ^^^
15:18:59 <ndevos> hagarth: fedora carries different .service units than upstream, and I think the ones from Fedora are better
15:19:00 <kkeithley> extras/systemd/glusterd.service
15:19:04 <hagarth> ndevos: ok
15:19:15 <kkeithley> should be the same in both
15:19:24 <hagarth> johnmark: the debian folks would like us to spellcheck our code base ;)
15:19:30 <kkeithley> if they're not....
15:19:34 <hagarth> that makes life easier for them to package
15:19:34 <ndevos> kkeithley: yes, but glusterfsd.service not yet, and glusterd.service needs some change too iirc
15:19:39 <kkeithley> correct
15:20:03 <hagarth> johnmark: apart from that, semiosis wanted a change for gf-errorcodes.h
15:20:20 <hagarth> johnmark: I don't think there are any other changes needed for .debs
15:20:49 <hagarth> ok, moving on to documentation for 3.5
15:20:49 <ndevos> but, I guess teh systemd changes can also wait - its been like this for a long time already...
15:20:52 <johnmark> hagarth: heh, ok
15:20:59 <hagarth> ndevos: right
15:22:40 <johnmark> oh, just noticed this - https://launchpad.net/ubuntu/trusty/+source/glusterfs/3.4.1-1ubuntu1
15:22:43 <johnmark> interesting
15:22:52 <johnmark> #topic documentation
15:23:08 <johnmark> yikes. netsplit?
15:23:50 * ndevos is still here
15:23:57 <Technicool> still here
15:24:06 <Technicool> i think just hagarth dropped
15:24:40 <johnmark> ah ok
15:24:44 <pk> he said he will be back
15:24:48 <pk> some network problem
15:24:49 <johnmark> pk: thanks :)
15:24:52 <johnmark> doh
15:25:04 <pk> :-)
15:25:22 <johnmark> semiosis: we're apparently finally getting into the mainline Ubuntu distro: https://launchpad.net/ubuntu/trusty/+source/glusterfs/3.4.1-1ubuntu1
15:25:29 <Technicool> wb hagarth
15:25:29 <hagarth> am back
15:25:43 <hagarth> Technicool: thanks and apologies for the incovenience
15:25:52 <hagarth> I was typing a monologue before I dropped off
15:25:55 <hagarth> let me paste it here now
15:26:03 <hagarth> the admin-guide can be browsed on github - https://github.com/gluster/glusterfs/tree/master/doc/admin-guide
15:26:09 <hagarth> however, content is not good in two ways:
15:26:14 <hagarth> 1. the content is not current
15:26:17 <johnmark> hagarth: ok
15:26:19 <hagarth> 2. content doesn't look good from a cosmetic point of view
15:26:25 <hagarth> we need some focus to clean that up
15:26:27 <johnmark> hagarth: what format?
15:26:32 <hagarth> the plan is to have some cleaning up done over the documentation hackathon weekend
15:26:32 * johnmark looks
15:26:37 <hagarth> as I mentioned earlier, I will create a howto for submitting documentation patches
15:26:48 <hagarth> johnmark: all of the source is in markdown format
15:27:20 <johnmark> hagarth: just saw that. Thanks
15:27:28 <hagarth> once we clean it up here to a certain degree, we need to roll it out on gluster.org
15:27:30 <johnmark> Technicool: can we get that into the docs repo on the forge?
15:27:33 <johnmark> hagarth: ok
15:27:45 <johnmark> would be good if we can unleash it at the same time as the new web site
15:27:54 <johnmark> but perhaps we should just move before then
15:27:55 <hagarth> johnmark: it already is in the glusterfs repo
15:28:05 <hagarth> ok, here's one question:
15:28:06 <Technicool> johnmark, yes but I am not seeing anything in the .md files?
15:28:21 <hagarth> should documentation be part of glusterfs repo or a separate repo of its own?
15:28:34 <johnmark> hagarth: that's a question we need to answer
15:28:38 <Technicool> hagarth, there is a seperate docs repo on the forge
15:28:40 <ndevos> I'd say separate
15:28:43 <johnmark> so there's the web site, which will have the docs "baked in"
15:29:06 <hagarth> my take is that it is easier to maintain documentation with new code changes that come in if it is part of the same repo
15:29:10 <johnmark> and the question is, should the docs be in a separate repo and then pulled into other repos when needed?
15:29:25 <hagarth> else, we'll need to make an effort to keep code and documentation in sync
15:29:38 <johnmark> hagarth: as long as the web site can pull from the docs dir whenever needed, I guess it doesn't matter where the docs live
15:29:40 <Technicool> i am looking at https://github.com/gluster/glusterfs/blob/master/doc/admin-guide/en-US/markdown/Administration_Guide.md and not seeing any content, which I am sure is due to the early time and lack of coffee but can someone sanity check me
15:30:01 <ndevos> I think the docs need to be kept in sync with the feature pages, adding the docs to the glusterfs repo itself sounds too complicated
15:30:25 <Technicool> johnmark, it would extra steps to the process of building the static site though
15:30:33 <ndevos> not all components that are mentioned in the docs are part of the glusterfs sources (samba plugin, swift, ...)
15:30:58 <johnmark> ndevos: why?
15:31:11 <johnmark> Technicool: would it? we need to figure out git hooks to automagically update that
15:31:15 <johnmark> ah crap
15:31:20 * johnmark looks for a git expert
15:31:48 <johnmark> because the workflow should be 1. docs edited/updated/added to software release 2. docs pulled into web site repo 3. docs published on web site
15:31:51 <Technicool> johnmark, it would but it might be minimal effort if it is just pulling from an additional repo
15:31:58 <hagarth> sorry about my network connectivity today, am on a flakey one.
15:32:16 <hagarth> did we get more opinions on documentation?
15:32:18 <johnmark> Technicool: ok. nad besides if the docs are in their own repo, we would still need to add the extra step of pulling/pushing into the web site content
15:32:35 <johnmark> 07:30 < ndevos> I think the docs need to be kept in sync with the feature pages, adding the docs to the
15:32:38 <johnmark> glusterfs repo itself sounds too complicated
15:32:41 <johnmark> 07:30 < johnmark> ndevos: why?
15:32:43 <johnmark> 07:30 < Technicool> johnmark, it would extra steps to the process of building the static site though
15:32:46 <johnmark> 07:30 -!- hagarth [~vijay@122.172.19.51] has quit [Read error: Connection reset by peer]
15:32:49 <ndevos> not all components that are mentioned in the docs are part of the glusterfs sources (samba plugin, swift, ...)
15:32:49 <johnmark> 07:30 < ndevos> not all components that are mentioned in the docs are part of the glusterfs sources
15:32:52 <johnmark> (samba plugin, swift, ...)
15:32:55 <johnmark> 07:31 < johnmark> Technicool: would it? we need to figure out git hooks to automagically update that
15:32:58 <johnmark> 07:31 < johnmark> ah crap
15:33:00 <johnmark> 07:31  * johnmark looks for a git expert
15:33:03 <johnmark> 07:31 < johnmark> because the workflow should be 1. docs edited/updated/added to software release 2.
15:33:06 <johnmark> docs pulled into web site repo 3. docs published on web site
15:33:09 <johnmark> 07:31 < Technicool> johnmark, it would but it might be minimal effort if it is just pulling from an
15:33:12 <johnmark> additional repo
15:33:13 <johnmark> ndevos: this is why we need to decide what constitutes a gluster release
15:33:17 <johnmark> hagarth: ^^^^ :)
15:33:19 <johnmark> ndevos: true
15:33:21 <hagarth> johnmark: thank you!
15:33:27 <johnmark> hagarth: sure
15:33:28 <johnmark> heh
15:33:51 <hagarth> should we arrive at a decision by putting out a question on the mailing list?
15:33:55 <johnmark> I had thought we could wait on deciding that until after the 3.5 release, but sounds like we need to do that now
15:34:12 <avati> the advantage of github is that you can concurrently maintain multiple release docs at no extra sync/cost
15:34:18 <johnmark> hagarth: sure - on gluster-devel. I think that's a question for maintainers of all the side projects + you and avati
15:34:31 <avati> github.com/gluster/glusterfs/tree/<version>/doc/...
15:34:42 <johnmark> avati: er, you can do that with any git repo
15:34:52 <avati> johnmark: not the MD rendering
15:34:59 <johnmark> avati: oh, that. ok
15:34:59 <hagarth> johnmark: ok, I will send out a mail on this one
15:35:00 <avati> others will dump raw .md files
15:35:18 <hagarth> #action hagarth to send out a mail on gluster-devel regarding repo for documentation
15:35:52 <hagarth> any other queries / suggestions on documentation?
15:36:00 * ndevos likes images and diagrams in documentation
15:36:04 <johnmark> ndevos: :)
15:36:18 <Technicool> i am assuming that the docs currently are not in sequential order?
15:36:21 <hagarth> ndevos: yeah, we need to get there :)
15:36:27 <hagarth> Technicool: as in?
15:37:16 <kkeithley> 27 8×10 color glossy pictures with circles and arrows and a paragraph on the back of each one explaining
15:37:24 <Technicool> e.g., admin_ACL's comes before admin_Hadoop, so when we are presenting the docs, are we presenting in the same order on the website or do we need to come up with some schema for how we order them?
15:37:45 <johnmark> kkeithley: :)
15:37:55 <johnmark> Technicool: I assume there's an intro/overview/index page somewhere...
15:37:59 <hagarth> Technicool: yeah, we can definitely do some re-structuring and re-ordering of content
15:38:06 <Technicool> 8x10 glossy - documentation glamour shots  ;)
15:38:24 <hagarth> ok, moving on to 3.4.2.
15:38:27 <hagarth> #topic 3.4.2
15:38:50 <hagarth> good news about 3.4.2 is that we have a few dht backports
15:39:08 <johnmark> hagarth: cool
15:39:31 <hagarth> thanks to shishir (who is not around today)
15:39:49 <hagarth> there are some regression failures and once that is sorted out, we can release 3.4.2
15:40:15 <hagarth> there have been requests to enhance CLI to o/p a warning when rdma and replace-brick are being used .
15:40:15 <johnmark> hagarth: ok, thanks
15:40:30 <hagarth> we need to get that in as well for 3.4.2.
15:40:41 <johnmark> hagarth: is there already a patch for that?
15:40:57 <hagarth> johnmark: not yet, any volunteers to send those patches to 3.4.2?
15:41:09 <johnmark> hagarth: how hard is it? :)
15:41:27 <hagarth> johnmark: it is pretty simple, if nobody gets there by Friday, I'll do it :)
15:41:38 <johnmark> hagarth: ok
15:41:52 <pk> johnmark: What is the bug-id?
15:42:00 <ndevos> hagarth: whats the link to the wishlist again?
15:42:04 <johnmark> pk: I don't know
15:42:07 <hagarth> pk - http://www.gluster.org/community/documentation/index.php/Backport_Wishlist
15:42:34 <hagarth> and another one here - #link https://bugzilla.redhat.com/show_bug.cgi?id=1039954
15:42:36 <glusterbot> Bug 1039954: medium, unspecified, ---, kaushal, NEW , replace-brick command should warn it is broken
15:43:17 <hagarth> I propose that we release 3.4.2 next week since these patches need to be in
15:43:26 <johnmark> hagarth: that sounds good
15:43:42 <johnmark> so looks like we'll have 3.5 beta by Friday, and then 3.4.2 by next week
15:43:50 <hagarth> #info 3.4.2 release planned for next week
15:44:00 <hagarth> johnmark: yes, that is the plan.
15:44:27 <hagarth> ok, let us move on.
15:44:39 <hagarth> #topic Support for released versions
15:45:19 * ndevos suggests to buy a Red Hat Storage subscription ;)
15:46:42 <johnmark> hagarth: updated backport wishlist so that descriptions are included with bug URLs
15:46:47 <johnmark> LOL
15:46:56 <johnmark> hagarth: you mean like an SLA?
15:47:40 <ndevos> auto closing bugs like Fedora does on End-Of-Life?
15:47:43 <johnmark> hagarth: or time limit on how long we'll support specicif versions?
15:47:54 <johnmark> hrm...l I think we lost him again :(
15:48:07 <hagarth1> johnmark: which was the last line you saw from me?
15:48:28 <kkeithley> (10:44:40 AM) hagarth: #topic Support for released versions
15:48:44 <hagarth> with 3.5 coming out, we will have an interesting situation
15:48:49 <hagarth> there will be at least three active release branches - 3.3, 3.4 and 3.5
15:48:53 <hagarth> how many of these branches can and should we support?
15:49:00 <johnmark> "yes that is the plan"
15:49:05 <hagarth> should we move to a time based model i.e. EOL support after 12 or 18 months?
15:49:12 <hagarth> or should we EOL support as releases keep happening?
15:49:19 <hagarth> i.e. we EOL 3.3 as soon as 3.5 happens?
15:49:28 <johnmark> hagarth: I think time-based makes sense
15:49:37 <johnmark> especially since we're moving to a time-based release schedule
15:49:40 <pk> hagarth: time-based
15:49:44 * kkeithley predicts a lot of moaning and wailing if we EOL 3.3
15:49:45 <avati> we need one release ($current) branch till the next release branch becomes "really stable"
15:49:51 <johnmark> kkeithley: agreed :(
15:49:52 <hagarth> johnmark: agree with that
15:49:58 <ndevos> +1 for release based - 3 stable releases?
15:49:58 <avati> *at least till
15:50:04 <johnmark> hagarth: we have to make sure there's enough overlap
15:50:29 <hagarth> kkeithley: yeah, some folks wanted to backport readdir-ahead to 3.3
15:50:30 <johnmark> ndevos: so with 3.5, we would still support 3.3 - 3.5 - that makes sense
15:50:35 <ira> Is 3.5 stable in beta?
15:50:44 <johnmark> ira: define "stable" :)
15:50:46 <avati> ira: i would say no
15:50:48 <hagarth> johnmark: ok, I think that makes sense.
15:50:51 <johnmark> ira: I think we're awaiting a couple more patches
15:50:52 <ira> For this definition.
15:50:53 <ndevos> johnmark: yes, and 3.3 would be EOL when 3.6 is entering beta/ga
15:51:02 <johnmark> ndevos: that sounds reasonable
15:51:14 <ira> So we'll maintain 4 branches? 3.3/3.4/3.5/master?
15:51:38 <kkeithley> only because 3.5 came so close behind 3.4
15:52:00 <hagarth> ira: in a sense yes. but the release branches are 3.3/3.4/3.5
15:52:05 <ndevos> wasnt the plan to release aprox. every 6 months?
15:52:09 <hagarth> master would be the active development branch
15:52:09 <abyss> johnmark: I sent an e-mail. Thank you.
15:52:10 <johnmark> ndevos: yes
15:52:21 <johnmark> abyss: awesome. you're welcome
15:52:29 <kkeithley> when we get back on 6 month cadence it'll only be 3.4/3.5/master,  3.5/3.6/master, and so on
15:52:38 <hagarth> ok, let us intend supporting 3 release trains for now.
15:52:59 <hagarth> everybody fine with that?
15:53:11 <avati> 2 release + master is reasonable
15:53:19 <avati> 3 release + master is a pain
15:53:25 <johnmark> hagarth: fine by me
15:53:33 <avati> (unless we have a volunteer dedicated for the 3rd release branch)
15:53:35 <pk> pk: agreed.
15:53:44 <johnmark> hagarth: says the guy who's not doing any dev work :)
15:53:58 <ndevos> 2 releases sounds good to me too
15:53:58 <johnmark> avati: yeah, we can ask for volunteers to maintain a 3rd branch
15:54:09 <johnmark> ira: howdy - welcome :)
15:54:20 <hagarth> ok, I haven't seen too many complaints with 3.3.2 as of now
15:54:35 <hagarth> if there are volunteers, we will release 3.3.3
15:54:45 <hagarth> else we will focus our attention on 3.4 and 3.5
15:55:05 <hagarth> ignore the else :)
15:55:29 <avati> :D
15:55:30 <hagarth> we will just focus our attentions on maintaining 3.4 and 3.5
15:55:35 <pk> hagarth: So if a release becomes eol, what is the story for backward compatibility?
15:56:01 <hagarth> pk: once a release is EOL, we can break compatibility between that release and the latest one
15:56:20 <pk> hagarth: Now you are talking :-)
15:56:31 <ndevos> lol!
15:56:36 <ira> hagarth: But you can go 3.3 -> 3.4 -> 3.5 -> 3.6 if you need to...
15:57:04 <avati> ira: i guess he means 3.3 client vs 3.6 server etc.
15:57:24 <ira> avati: On disk, also was a question...
15:57:34 <hagarth> ira: sure, that would be the upgrade path. I was referring to client-server compatibility.
15:57:38 <kkeithley> so far we think that works, but I think we're talking about what we _guarantee_, yes?
15:57:41 <ira> hagarth: Ah, ok.
15:58:01 <avati> ira: we haven't made any on-disk incompatibility changes (yet)
15:58:48 <hagarth> kkeithley: yes, until we get a dedicated testing unit in place, it is most likely to be best effort and guaranteeing is going to be rather difficult.
15:58:51 <pk> avati: Why not also talk about on-disk compatibility guarantees for eol releases, mainly xattrs
15:59:32 <ira> pk: I think because it hasn't happened.  I'm asking because I've watched projects deal with it... so I'm just tossing it out to be thought on.
15:59:38 <hagarth> pk: if we were to make any on-disk changes, we would need to provide a mechanism to upgrade/clean up.
16:00:07 <pk> hagarth: makes sense.
16:00:34 <hagarth> we are moving close to the end of the scheduled time. I think we have a fair agreement on our support stance for releases.
16:00:40 <hagarth> we can discuss backward compatibility in more detail next week.
16:00:43 <hagarth> #topic open discussion
16:01:08 <hagarth> anything up for discussion here?
16:01:14 <kkeithley> FYI, I'm on PTO starting tomorrow through the EOY.
16:01:27 <kkeithley> I'll check email, mainly so I don't have 9000000 emails in my inbox
16:01:42 <avati> I'm on vacation next week (w/ adjacent weekends)
16:01:48 <kkeithley> and I'll build RPMs if I can, unless ndevos beats me to it
16:01:50 <hagarth> kkeithley: have a good vacation!
16:01:57 <ira> kkeithley, avati : Have a good vacation! :)
16:01:58 <johnmark> kkeithley: have fun :)
16:02:00 <kkeithley> thanks
16:02:01 <johnmark> what' svacation?
16:02:02 <ndevos> oh, I'm happy to do so kkeithley
16:02:06 <hagarth> avati: ok
16:02:11 <johnmark> hagarth: thank you
16:02:25 <hagarth> ending today's meeting
16:02:29 <hagarth> #endmeeting