15:00:34 <hagarth> #startmeeting
15:00:34 <zodbot> Meeting started Wed Jan 22 15:00:34 2014 UTC.  The chair is hagarth. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:34 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
15:00:35 <glustermeetbot> Meeting started Wed Jan 22 15:10:28 2014 UTC.  The chair is hagarth. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:35 <glustermeetbot> Useful Commands: #action #agreed #help #info #idea #link #topic.
15:00:50 <jdarcy> Good morning.
15:01:10 <hagarth> who do we have here today?
15:01:22 * vbhat is here
15:01:24 <Technicool> morning
15:01:34 <ira> Hello.
15:01:40 * foster 
15:01:43 * lalatenduM here
15:01:44 * jdarcy is present.
15:01:50 <raghu`> hello
15:02:02 <hagarth> cool, we seem to have a high degree of quorum today :)
15:02:06 <hagarth> let us get rolling.
15:02:16 <hagarth> #topic AI follow up
15:02:38 <hagarth> we had 2 items from the previous meeting
15:02:39 * kkeithley is here
15:03:21 <hagarth> and both seem to have been covered. lalatenduM and me are having a conversation with johnmark on Hyperkitty and will update you when we have more concrete details.
15:03:39 <hagarth> moving on
15:03:40 <hagarth> #topic 3.5.0
15:03:53 <kkeithley> wait, how about the AI from two weeks ago
15:04:07 <hagarth> kkeithley: which one was that?
15:04:10 <kkeithley> you and me syncing up on 3.3.3 and 3.4.3
15:04:24 <kkeithley> commit bit in gerrit, etc.
15:04:58 <kkeithley> whatever else I need
15:05:10 <kkeithley> for release wrangling
15:05:11 <hagarth> kkeithley: ok, I thought we had it covered. I will initiate a discussion later this week with you.
15:05:23 <kkeithley> okay
15:05:43 <hagarth> #action kkeithley and hagarth to meet later this week on release wrangling.
15:05:49 <hagarth> okay, now on to 3.5.0
15:05:57 <hagarth> the first beta got out last week
15:06:09 <hagarth> and we had some bugs logged during the test weekend
15:06:31 <hagarth> I have noticed a few bugs with compression xlator as well (some crashes and problems in working with afr)
15:06:48 <hagarth> I am yet to convert those to bugs .. will do so and update the 3.5.0 tracker
15:07:23 <hagarth> there has been a request to change the name of the key "compression" in volume set for enabling on-wire compression
15:07:37 <hagarth> any opinions on that?
15:07:49 <jdarcy> What was the rationale for that?
15:07:54 <hagarth> #link https://bugzilla.redhat.com/show_bug.cgi?id=1053670
15:08:00 <glusterbot> Bug 1053670: high, unspecified, ---, kaushal, NEW , "compress" option name for over-wire-compression is extremely misleading and should be changed
15:08:13 <hagarth> jdarcy: some folks feel that the name "compression" refers to compression at rest
15:08:41 <jdarcy> It *would* be nice to distinguish between the two types of compression IMO.
15:09:03 <hagarth> agree, should we change this to network-compression or something like that?
15:09:09 * vbhat thinks volume set "compression" should be changed to volume set "compress-over-wire" or something similar
15:09:31 <hagarth> and have disk-compression when we evolve at rest compression xlator?
15:09:32 <jdarcy> "Network" vs. "volume" or some such, yeah.
15:09:36 <ira> And do we note where the translator is in the stack?  (Does compression imply over the wire, or not?)
15:09:50 <ndevos> yeah, it would be good to have the disticion
15:10:07 * ndevos wonders how Wireshark should handle the compression...
15:10:14 <hagarth> ira: the translator is loaded in both client and server stacks
15:10:38 <ira> hagarth: But if we are doing file compression... it is valid to offload onto the client?
15:10:46 <hagarth> ndevos: the headers should not get mangled.
15:11:03 <hagarth> ira: no, this is just payload compression for network transmission
15:11:11 <johnmark> gah
15:11:18 <johnmark> sorry, guys. hotel wifi :(
15:11:21 <ndevos> hagarth: ah, thats a relief, so only read/write data, or also iatt's and other structures?
15:11:24 <ira> Understood, I was speaking about the on the disk type .
15:11:47 <hagarth> ndevos: IIRC, only read/write.
15:12:08 <hagarth> ira: we do not yet have a on disk compression xlator.
15:12:12 <ndevos> hagarth: okay, then I'm fine :)
15:12:33 <hagarth> ndevos: it should be fun to play with compression + encryption + wireshark :)
15:12:49 <ndevos> hagarth: no, it is not.
15:13:06 <hagarth> should we vote here on the new name for the option or follow up with a ML discussion?
15:13:26 <jdarcy> I vote for a Gerrit discussion.  ;)
15:13:35 <lalatenduM> jdarcy, +1
15:13:51 <hagarth> jdarcy: sounds like a good idea :)
15:13:51 <ndevos> gerrit +1
15:13:58 <johnmark> hagarth: if we can limit the choices to some small number, we could ask the dev list
15:14:18 <johnmark> but I don't think it's terribly important - make a choice and stick to it
15:14:40 <hagarth> gerrit seems to be the most preferred option. Let us do it there.
15:14:57 <johnmark> cool
15:15:02 <hagarth> #action new key for option compression to be done on gerrit
15:15:29 <hagarth> coming to other features in beta1
15:15:38 <jdarcy> Who has that AI?  Whoever it is should probably refer to these minutes.
15:15:47 <hagarth> quota needs some patches and raghu` has sent across a few on master now
15:16:17 <hagarth> any takers for driving a new key/name for compression?
15:16:41 * ndevos votes for jclift as he filed the bug :)
15:16:57 <jdarcy> Bug is assigned to Kaushal, FWIW.
15:16:57 <hagarth> ndevos: that does look like a great idea!
15:16:58 <kkeithley> driving it? I vote for calling it wire-compression
15:17:17 <ndevos> kkeithley: network.compression!
15:17:23 <kkeithley> how much discussion or driving does it need.
15:17:29 <kkeithley> okay, network.compression
15:17:34 <kkeithley> sold
15:17:47 <hagarth> ok, let us await comments on gerrit with this option :)
15:17:57 <hagarth> coming back to 3.5.0
15:18:00 <ndevos> I can file a patch for that
15:18:11 <hagarth> ndevos: thanks, you own the AI now ;)
15:18:26 <hagarth> geo-replication patches are also trickling in
15:18:44 <hagarth> so we should be able to do a refresh of the beta this week after the quota patches get merged
15:19:04 <lalatenduM> hagarth, sounds good
15:19:22 <hagarth> we do need some more testing for the new content in 3.5.0
15:19:40 <hagarth> johnmark: do we plan to have one more test day/weekend this week?
15:19:49 <johnmark> hagarth: yes.
15:20:02 <johnmark> if we know beta 2 is coming out by tomorrow, then testing weekend it is
15:20:05 <vbhat> When's the next test day?
15:20:11 <johnmark> vbhat: this weekend
15:20:13 <hagarth> johnmark: cool, we definitely need some more testing for 3.5.
15:20:16 <johnmark> so Friday - Monday
15:20:27 <johnmark> hagarth: when can we expect beta 2?
15:20:40 <hagarth> johnmark: I will shoot for tomorrow
15:20:43 <johnmark> I can make sure there's an announcement for the testing weekend today
15:21:00 <johnmark> hagarth: cool. I do need help from some of you here to drive the testing agenda
15:21:08 <johnmark> hrm.. Ben isn't here
15:21:08 <jdarcy> 3.5 bug list: https://bugzilla.redhat.com/buglist.cgi?bug_status=NEW&bug_status=ASSIGNED&classification=Community&list_id=2127619&order=Importance&product=GlusterFS&query_format=advanced&version=3.5.0
15:21:27 <hagarth> johnmark: agree
15:21:31 * johnmark looks at lalatenduM :)
15:21:44 <hagarth> johnmark: should we have a test week (i.e whole of next week) instead of a weekend?
15:22:12 <lalatenduM> johnmark, lets catch up and see what I can do :)
15:22:33 <lalatenduM> testweek seems to be good idea
15:22:48 <johnmark> hagarth: not a bad idea
15:22:58 <lalatenduM> testweek +1 as this is beta2
15:23:01 <johnmark> lalatenduM: cool
15:23:16 <hagarth> any ideas on getting more test coverage?
15:23:17 <johnmark> so we could go from Friday to Friday
15:23:32 <hagarth> johnmark: +1
15:23:33 <johnmark> hagarth: I think the first tests need to come from the dev team / QE
15:23:44 <johnmark> hagarth: and I need someone to review what I posted at gluster.org/gfest
15:24:06 <lalatenduM> johnmark, i have gone trough gluster.org/gfest, it is fine
15:24:12 <johnmark> lalatenduM: thanks :)
15:24:22 <lalatenduM> s/trough/through/
15:24:28 <vbhat> johnmark, Are we covering new geo-rep there?
15:24:28 <hagarth> johnmark: right, probably we need to have a working group for this test week.
15:24:29 <johnmark> but yeah, the basic point is that I wanted to divide testing into two basic categories
15:24:36 <johnmark> vbhat: only if it's in the beta
15:24:51 <johnmark> testing for features and, new for this release, performance testing
15:25:02 <johnmark> and I was hoping we could use Ben England's smallfiles project for that
15:25:04 <vbhat> johnmark, This is going to be in beta2. distributed-geo-rep
15:25:13 <johnmark> vbhat: then test it, we shall :)
15:25:15 <hagarth> johnmark: will review the gfest doc later.
15:25:20 <johnmark> hagarth: thanks
15:25:50 <johnmark> #action jmw to send out glusterfest announcement today
15:25:57 <hagarth> johnmark: should we form a small working group for this test week?
15:26:06 <johnmark> hagarth: excellent idea
15:26:22 <johnmark> hagarth: we can have people who attend here, as well as send an invite to the lists
15:26:25 <johnmark> both lists
15:26:37 <hagarth> johnmark: great, count me into that working group ;)
15:26:39 <lalatenduM> hagarth, what will be role or function of this test group?
15:27:19 <hagarth> lalatenduM: to divide tests across functional areas and help with co-ordinating the test fest.
15:27:22 <johnmark> hagarth: woot
15:27:59 <hagarth> lalatenduM: s/divide/divide and run/
15:28:19 <vbhat> hagarth, Cool.. I can help with geo-rep test cases
15:28:32 <hagarth> vbhat: awesome
15:28:42 <lalatenduM> hagarth, cool , I can run some re-balance tests
15:28:55 <hagarth> johnmark: we probably should capture the working group details in a wiki page.
15:29:18 <johnmark> hagarth: +1
15:29:27 <lalatenduM> hagarth, and invite people to join the group :0
15:29:31 <lalatenduM> :)
15:29:32 <johnmark> #action jmw to set up testing working group page
15:29:43 <johnmark> heh
15:29:48 <hagarth> lalatenduM: of course, johnmark will help in driving membership :)
15:30:02 <johnmark> lol
15:30:17 <johnmark> yup, because it makes my job easier :)
15:30:33 <hagarth> ok, that was it on 3.5. Any questions?
15:30:51 <hagarth> moving on
15:31:06 <hagarth> #topic 3.4
15:31:18 <hagarth> kkeithley: any updates on 3.4 that we want to share here?
15:32:16 <hagarth> the backport wishlist #link http://www.gluster.org/community/documentation/index.php/Backport_Wishlist is empty
15:32:17 <kkeithley> no, nobody has added any backport requests
15:32:33 <hagarth> hmm does that mean everybody is very happy with 3.4.2? :)
15:32:37 <kkeithley> I just posted reminders on #gluster and #gluster-devel.
15:32:38 <kkeithley> I guess
15:32:50 <kkeithley> Will send reminders to both lists in a moment
15:33:06 <kkeithley> Might be the easiest job I was ever volunteered to do
15:33:15 <hagarth> kkeithley: sure, I thought I saw a bug report on not being able to change loglevel through volume set
15:33:35 <hagarth> If it is a real issue, it would be good to include that in 3.4.3.
15:33:45 <kkeithley> yes, and I believe I've seen some reviews go through gerrit
15:33:48 <kkeithley> for 3.4
15:34:19 <hagarth> kkeithley: right, we can review the backlog over the meeting that we intend having.
15:34:31 <hagarth> I think we need to fix https://bugzilla.redhat.com/show_bug.cgi?id=976750 for 3.4.3
15:34:37 <glusterbot> Bug 976750: low, medium, ---, vagarwal, CLOSED CURRENTRELEASE, Disabling NFS causes E level errors in nfs.log.
15:35:26 <hagarth> I have observed this annoying behavior when I have disabled nfs during my tests
15:35:54 <hagarth> anything else on 3.4?
15:36:21 <hagarth> figure not, moving on
15:36:21 <hagarth> #topic 3.6
15:36:37 <hagarth> we got the first feature proposal for 3.6 in the planning page this week
15:36:40 <jdarcy> I still need to add the different-size-brick feature request from the ML.
15:36:51 <hagarth> jdarcy: right
15:37:08 <hagarth> we also need to carry forward some features from the 3.5 planning page too.
15:37:19 <jdarcy> I was just typing the same thing.
15:37:38 <johnmark> hagarth: right
15:37:40 <jdarcy> SSL, certainly.  Someone was just asking about that.
15:38:15 <lalatenduM> SSL+1
15:38:18 <lalatenduM> SSL +1
15:38:31 <hagarth> jdarcy: yeah, and my object count work is mostly complete. Will rebase on top of latest quota and send it across.
15:38:56 <hagarth> I feel that we should do data-classification for 3.6.
15:39:28 <lalatenduM> hagarth, it is an excellent idea :)
15:39:42 <ndevos> I've seen an issue with users that are in *many* groups, anyone thougth about server-side-gid checking? (it's available for NFS, not for the native protocol)
15:39:43 <jdarcy> I'd love to, but that's probably subject to me getting free from NSR/1000-node/prog-guide.
15:40:15 <hagarth> btw, I am working on a trello board to capture backlog for GlusterFS. It is in a primitive state as yet but do let me know if you are interested in observing/collaborating on that board.
15:40:56 <jdarcy> Definitely interested.
15:41:14 <hagarth> jdarcy: I am very inclined to pick up data-classification if we don't find any takers.
15:41:28 <ira> hagarth: I'd be interested in reading it.
15:41:28 <jdarcy> hagarth: That would be magnificent.
15:41:46 <lpabon> hagarth, what is meant by data-classification?
15:41:51 <hagarth> ira, jdarcy: I will add you folks on to trello.
15:41:57 <kkeithley> me too please
15:42:09 <hagarth> lpabon: #link http://www.gluster.org/community/documentation/index.php/Features/data-classification
15:42:14 <hagarth> kkeithley: sure
15:42:25 <lpabon> hagarth, cool thanks
15:42:30 <raghu`> hagarth: add me also to trello board
15:43:08 <hagarth> for those who are reading the logs: if you do not have a trello account, please sign up on trello.com and send me a note.
15:43:12 <hagarth> raghu`: will do
15:43:18 * lalatenduM thinks data-classification is storage tiering
15:43:23 <vbhat> hagarth, Please add me as well
15:43:28 <hagarth> lalatenduM: yes
15:43:41 <hagarth> vbhat: noted
15:44:01 <jdarcy> lalatenduM: It's more than that.  Also e.g. rack-aware placement.
15:44:29 <lalatenduM> jdarcy, awesome :)
15:44:50 <hagarth> please do not keep your 3.6 feature page additions  for the last week.. proposing something early on will help in better reviews
15:45:30 <hagarth> ok, that seems to be all I had on 3.6. Any questions?
15:45:58 <johnmark> hagarth: thanks
15:46:03 <hagarth> figure not, moving on
15:46:10 <johnmark> remember, if it's not on a feature page, it doesn't exist :)
15:46:31 <hagarth> johnmark: right :)
15:46:36 <hagarth> #topic Bug triaging
15:46:53 <hagarth> we have a considerable bugzilla backlog now
15:47:04 <hagarth> some bugs do not get attended to
15:47:11 <kkeithley> https://bugzilla.redhat.com/show_bug.cgi?id=1056085
15:47:13 <glusterbot> Bug 1056085: unspecified, unspecified, ---, rgowdapp, NEW , logs flooded with invalid argument errors with quota enabled
15:47:30 <hagarth> I have been thinking about how we can get better with respect to triaging.
15:48:12 <hagarth> should we have a list of volunteers and take on bug triaging for a certain time slice?
15:48:29 <hagarth> s/and/and each takes/
15:48:53 <hagarth> maybe something like 2 weeks or so?
15:49:22 <hagarth> any other ideas for managing bug backlog/triaging of inflow?
15:49:56 <johnmark> hagarth: are any of the QE people monitoring that?
15:50:18 <hagarth> johnmark: not that I am aware of
15:50:38 <lalatenduM> hagarth, I want to volunteer , but I am very new to triaging bugs , if somebody can help me out
15:50:45 <hagarth> fwiw, we have 584 open bugs in bugzilla.
15:50:48 <johnmark> lalatenduM: you are my new favorite person
15:51:22 <hagarth> lalatenduM: I can help you out and we can evolve a document on what triaging involves.
15:51:28 <johnmark> lalatenduM: I think it would involve sending requests to developers to investigate
15:51:36 <johnmark> and then constantly reminding them :)
15:51:50 <lalatenduM> hagarth, johnmark , cool
15:51:58 <johnmark> until they get so annoyed that they have no choice but to fix it
15:52:03 <hagarth> johnmark: to start with, setting the appropriate severity/priority would be helpful :)
15:52:07 <lalatenduM> lol :0
15:52:14 <ira> IMHO: Triage, and follow up are two different things ;)
15:52:37 <ira> Triage -> Backlog -> Actioned.
15:52:38 <hagarth> ira: right, triage should help in identifying the set of bugs/folks that need to be followed up.
15:52:44 <johnmark> ira: 'tis true :)
15:53:07 <jdarcy> Do we have agreed-upon definitions for priority and (especially) severity yet?
15:53:17 <hagarth> I am a bit concerned that we might be losing something valuable by not triaging effectively.
15:53:37 <lalatenduM> hagarth, agree
15:53:40 <ndevos> lalatenduM: while you are at that, you'll figure out what components belong to which devs and you can update the MAINTAINERS file :)
15:53:41 <ira> hagarth: I'll almost guarantee we are? ;)
15:53:57 <lalatenduM> ndevos, yup, perfect :)
15:54:07 <hagarth> jdarcy: priority is what developers look at and severity helps in determining the order of bugs in the same priority queue.
15:54:27 <hagarth> ira: you bet ;)
15:54:36 <ira> What does priority mean though?
15:54:52 <ira> What is priority X vs Y is the question...
15:55:06 <jdarcy> hagarth: What I meant was, do we agree on what constitutes "urgent" vs. "high" vs. "medium" vs. "low" severity?
15:55:17 <ira> Same with sev.
15:55:26 <hagarth> jdarcy: ah okay, we probably should define that in the document that we evolve.
15:55:38 <hagarth> ira: right
15:55:43 <jdarcy> As I see it, severity is a technical measure that then feeds into priority which is a project decision.
15:55:44 <ira> You may want to start with some definitions.... and see how they work.
15:55:57 <jdarcy> Just getting severity set would be a start.
15:56:08 <ira> jdarcy: I'd agree with that.
15:56:12 <hagarth> #action lalatenduM and hagarth to send out a write up on bugzilla workflow/ triaging.
15:56:34 <ndevos> https://bugzilla.redhat.com/page.cgi?id=fields.html#priority ?
15:56:34 <hagarth> jdarcy: right
15:57:30 <lalatenduM> ndevos, thanks for the link, you should it with #link :)
15:57:34 <hagarth> anything else on bug triage?
15:57:38 <ndevos> #link https://bugzilla.redhat.com/page.cgi?id=fields.html#priority
15:57:42 <johnmark> also, could I make a recommendation that we look at how bugs are triaged in the samba, openstack and perhaps other projects?
15:57:50 <jdarcy> ndevos: There's also a storage-specific set of definitions, e.g. "urgent" is something that might compromise data vs. "high" only affects availability (such as a crash).
15:58:15 <lalatenduM> johnmark, I agree
15:58:26 <hagarth> johnmark: right, I am aware of how it happens in openstack. ira probably can chip in on samba. maybe a good discussion for next week?
15:58:27 <ndevos> jdarcy: yes, sure, but this is the bugzilla definition that gets showed when you click on 'Priority'
15:58:39 <ndevos> showed!? shown!
15:58:41 <ira> johnmark: In rhs-smb, I'm all in favor.  Samba has its own internal processes, and you may not want to go there ;)
15:58:59 <johnmark> haha ok :)
15:59:09 <hagarth> ira: ok :)
15:59:19 <hagarth> #topic open discussion
15:59:51 <ndevos> nightly builds are running: http://download.gluster.org/pub/gluster/glusterfs/nightly/
15:59:53 <jdarcy> Summit schedule is being formed as we speak.
15:59:57 <lalatenduM> hagarth, we need to fix github for glusterfs
15:59:58 <lalatenduM> :)
16:00:07 <hagarth> lalatenduM: thanks for the reminder :)
16:00:16 <hagarth> ndevos: awesome, thanks!
16:00:27 <lalatenduM> for context , github now a days doesnot have latest code
16:00:43 <ndevos> hagarth: it still needs some documentation, that being worked on...
16:00:46 <lalatenduM> I think the replication is broken , as mentioned by hagarth
16:00:50 <hagarth> jdarcy: I see, we sure should have a few talks on glusterfs in the summit.
16:01:13 <hagarth> lalatenduM: I will definitely investigate more on mirroring - both to forge and github.
16:01:26 <hagarth> according to me, gerrit is broken - not just its replication ;)
16:01:26 <lalatenduM> I have another question
16:01:33 <lalatenduM> hagarth, got it :)
16:01:33 <hagarth> lalatenduM: go ahead
16:01:55 <lalatenduM> is there any plan to publish the admin guide as html pages?
16:02:12 <lalatenduM> we have markdown copies in the code repo
16:02:17 <hagarth> lalatenduM: yes, that is the plan. we need some attention to fix the admin guide source in markdown.
16:02:36 <lalatenduM> it will help users to get good pointers through search engines
16:02:37 <jdarcy> johnmark: Do we have any lab/hangout/whatever plans for Summit?
16:02:39 <hagarth> once that happens, we can evolve a process to publish admin guide in html.
16:02:47 <lalatenduM> hagarth, ok
16:02:56 <johnmark> jdarcy: not yet
16:03:09 <johnmark> jdarcy: there will be an open area like last year
16:03:23 <johnmark> jdarcy: but yes, we should start planning what else we want to do
16:04:07 <hagarth> johnmark: have some thoughts reg. summit. let us plan it out.
16:04:12 <johnmark> perhaps that can be a topic for next week
16:04:17 <jdarcy> I could bring some shirts.  http://www.customink.com/signup/53rpwctm
16:04:18 <johnmark> hagarth: nice - yes, let's
16:04:22 <johnmark> heh
16:04:24 <hagarth> johnmark: right
16:04:38 <johnmark> ok, got to move to the next meeting
16:04:39 <hagarth> jdarcy: +1 L for me ;)
16:04:41 <johnmark> thanks, guys!
16:04:48 <hagarth> thanks everyone!
16:04:53 <hagarth> #endmeeting