bigdatasig
LOGS
14:01:52 <tstclair> #startmeeting
14:01:52 <zodbot> Meeting started Thu Jun 19 14:01:52 2014 UTC.  The chair is tstclair. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:01:52 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
14:02:01 <tstclair> #meetingname bigdatasig
14:02:01 <zodbot> The meeting name has been set to 'bigdatasig'
14:02:17 <tstclair> #meetingtopic status
14:03:07 <tstclair> Hey folks, want to give quick update?  pmackinn rsquared mizdebsk willb_
14:03:15 <tstclair> Lots of stuff is broken atm.
14:03:44 <willb_> yeah, my main status is that I need to fix some bugs in Fedora scala packages and then wait for rawhide churn to settle down before fixing everything else
14:04:21 <rsquared> latest version of hbase almost done.  Snag on the latest rebase and waiting on guava update.  Direction for hadoop is understood, but there's a 2.4.1 release I will probably jump to when guava revs
14:05:20 <pmackinn> status: adding datanucleus-maven-parent pkg for dn upgrades to 3.x latest; tying to work around system scope restrictions in xmvn 2 for dn pkgs
14:05:23 <mizdebsk> there were some changes in maven environment (javapackages and xmvn updates), but most build failures should be easy-fix
14:05:41 <mizdebsk> guava update will be next week
14:06:24 <tstclair> mizdebsk, rsquared ^ do we have an impact graph?
14:06:33 <tstclair> were 1 month from alpha
14:07:12 <rsquared> tstclair: No idea, but guava isn't shy about removing apis so I'd guess it could be a non-trivial impact
14:07:47 <mizdebsk> graph no, direct dependants of guava are there: https://lists.fedoraproject.org/pipermail/java-devel/2014-June/005274.html
14:08:28 <mizdebsk> maven requires guava so graph would contain basically everything :)
14:08:37 <number80> hello
14:08:44 <tstclair> willb_, ^ spark
14:08:52 <tstclair> +solr
14:09:05 <tstclair> pmackinn, ambari
14:09:11 <tstclair> I've got bookkeeper.
14:09:12 <pmackinn> shirley
14:09:21 <rsquared> hadoop and hbase are on the list.  Possible oozie
14:09:49 * pmackinn expects turbulence at this point
14:10:17 <tstclair> I'm not even going to try fixing my packages until the churn settles.
14:10:32 <tstclair> Too many other tasks in flight.
14:10:53 <tstclair> mizdebsk, what's the eta?
14:11:08 <mizdebsk> i could delay the update 1 week more and provide copr repo, but i don't expect people to even try that
14:11:38 <mizdebsk> tstclair: i have update mostly ready, will push it on mon or tup
14:12:05 <pmackinn> mizdebsk, summer is here...have vacation plans yet?
14:12:14 <rsquared> With alpha this close I'd rather guava just land and we'll deal with it
14:12:14 <tstclair> ;-)
14:12:27 <tstclair> +1 just land it.
14:12:54 <mizdebsk> no plans yet
14:13:19 <tstclair> mizdebsk, are there other shifts that we should be aware of in the pipe?
14:14:09 <mizdebsk> maven 3.2.2 is already built in koji but not tagged
14:14:23 <mizdebsk> i expect to tag it within a few days
14:15:13 <mizdebsk> soon all packages using jpackage_script will connect to abrt, if installed
14:16:30 <mizdebsk> nothing more atm
14:18:08 <tstclair> So I'm planning on hold off updates util everything lands.  May be 1st week into July
14:19:18 <tstclair> number80, do you have an ETA of when you want things stable prior to cloud images?
14:19:49 <rsquared> mizdebsk: Anything expected to break from the maven 3.2.2 update?
14:20:24 <mizdebsk> rsquared: something always breaks from maven updates :)
14:20:36 <mizdebsk> but there shouldn't be anything big
14:21:03 <number80> tstclair: I think we should focus on having a core hadoop + stable components but having some tech preview should be fine
14:21:29 <tstclair> number80, yup, but did you have a potential timeframe?
14:21:40 <number80> I'd like to have by the end of the month a list of the components we should ship in the image
14:21:47 <number80> (sorry, still slow typing mode)
14:22:03 <number80> alpha is july, 22
14:22:21 <tstclair> We'll try to enumerate the list of images, by next week.
14:22:30 <tstclair> I'll take that as an action.
14:22:32 <number80> we should have at that point a preliminary image and then focus on stabilizing them
14:22:35 <number80> tstclair: thanks
14:22:43 <rsquared> number80: I'll be spinning new hbase and hadoop packages next week I hope.  oozie hopefully soon after.
14:22:51 <number80> rsquared: great :)
14:23:15 <tstclair> number80, I would normally agree ^ but as you can see stabilizing may take a bit.
14:23:30 <pmackinn> number80, any update on sqoop review? or are you blocked?
14:23:31 <rsquared> And I'm gone for 1 1/2 weeks at the end of june/early july
14:23:34 <number80> btw, we're actually discussing of having a joint cloud images testing days around july, 24
14:24:04 <tstclair> number80, good to know.
14:24:09 <number80> pmackinn: still that unowned stuff, but as soon as it's fixed, i'll validate it
14:24:31 <number80> pmackinn: i'll update the ticket today if that's blocking you (my apologies)
14:25:26 <pmackinn> number80, yeah i should rebuild with xmvn 2...thanks
14:28:05 <number80> good
14:29:05 <tstclair> k, are there any other issues to discuss?
14:29:31 <tstclair> number80, mizdebsk, rsquared, pmackinn, willb_ ?
14:29:33 <mizdebsk> do we have or are we going to prepare test plan(s) at least for the components that have f21 features accepted?
14:29:56 <mizdebsk> this could be a good way to let people familiarize with big data and at the same time help testing
14:30:10 <willb_> mizdebsk, that's a great question
14:30:17 <tstclair> mizdebsk, good point.
14:30:49 <number80> roshi wants to jump in about testing
14:31:22 <number80> hi
14:31:23 <tstclair> greetings roshi.
14:31:29 <roshi> hello all :)
14:31:32 <roshi> o/
14:31:34 <number80> for roshi:
14:31:37 <tstclair> mizdebsk, was talking about test plans.
14:31:50 <number80> mizdebsk | do we have or are we going to prepare test plan(s) at least for the components that have f21 features accepted?
14:31:57 <number80> mizdebsk | this could be a good way to let people familiarize with big data and at the same time help testing
14:32:02 <number80> eof
14:32:56 <tstclair> number80, what's the deliverable and to who?
14:33:12 <tstclair> test plan = what, hand off - to who, by - when?
14:34:18 <number80> deliverable should be an image that could be used to deploy hadoop + ecosystem on single/multi-node status
14:34:36 <number80> if we can't do that, having a master and slave images is ok
14:34:42 <number80> (in time)
14:35:41 <number80> discussion is open about deliverables but from our POV, we should focus on having a functional images for F21 and then improving if needed
14:36:10 <tstclair> I think handing off blog posts is probably the most logistically possible at this point, given the breakage.  If there are folks who are interested, we can point to the posts.
14:36:23 <number80> +1
14:36:31 <tstclair> Should we send emails to -devel -test ?  which list?
14:36:55 <number80> I think roshi could answer this more properly
14:37:07 <number80> (it would be nice to add -cloud I think)
14:37:14 <tstclair> roshi, ^?
14:37:55 <roshi> as for deliverables for a test day
14:38:46 <roshi> test cases you want to run, tutorials for people to get started (unless you're expecting all participants to already have a certain base knowledge) and blog posts to publicize it in your channels
14:39:00 <roshi> test@ and cloud@ would be good
14:39:27 <tstclair> ack.
14:39:47 <roshi> I can help with any pointers as we move forward
14:39:59 <number80> so the test cases could be the blog posts
14:40:13 <number80> do you think we should appoint a facilitator ?
14:40:21 <roshi> they could start that way - but then we'd move them to the wiki with the rest of the test cases
14:40:28 <number80> ok
14:40:40 <roshi> especially if there were things that would be good *in general* to test, not just on test days
14:40:40 <tstclair> So todo list = "list of images to create w/packages + blog posts to email lists"
14:41:10 <roshi> it generally helps if you tag someone to handle it
14:41:22 <roshi> having a single POC makes stuff easier as a whole, I think
14:42:06 <roshi> http://fedoraproject.org/wiki/QA/SOP_Test_Day_management
14:42:23 <roshi> that's the QA guide to test days - and basically what we'll be working off
14:42:49 <roshi> I'll help whoever you guys tag as the facilitator get everything ironed out
14:44:14 <roshi> if that's good for you guys, I'll stop derailing the meeting :p
14:44:53 <number80> roshi: you're more than welcome here :)
14:45:27 <roshi> thanks :)
14:45:44 <tstclair> k, round 2: mizdebsk, number80, roshi, rsquared, pmackinn, willb_  - any other items?
14:45:58 <rsquared> Nothing from me
14:45:59 <willb_> none here
14:46:13 * roshi has nothing :)
14:46:28 <number80> nope
14:46:41 <mizdebsk> just info/request about ivy
14:46:59 <mizdebsk> ivy packaging improvements were implemented, but there is no documentation yet
14:47:06 <mizdebsk> without help from others i can't be really sure how well it works
14:47:15 <mizdebsk> if there are people that want to use improved ivy or sbt - please get in touch with me
14:47:16 <tstclair> willb_, ^
14:47:23 <mizdebsk> i would like to test some packages "in production" by updating to new packaging way and pushing to rawhide
14:47:34 <willb_> mizdebsk, awesome, let's chat
14:47:58 <mizdebsk> that's it from me
14:48:03 <mizdebsk> i have no more items
14:48:06 <willb_> I'll figure out which of my sbt packages are easiest and toughest ;-)
14:48:22 <tstclair> ack, thanks folks.
14:48:33 <tstclair> #endmeeting