bigdata_sig
LOGS
15:01:57 <tstclair> #startmeeting
15:01:57 <zodbot> Meeting started Thu Jan 16 15:01:57 2014 UTC.  The chair is tstclair. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:01:57 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
15:02:14 <tstclair> #meetingname BigData_SIG
15:02:14 <zodbot> The meeting name has been set to 'bigdata_sig'
15:02:42 <tstclair> #meetingtopic status
15:02:56 <tstclair> #topic open_issues
15:03:39 <tstclair> Morning folks, before we roll onto status, are there any open issues that exist?
15:04:07 <tstclair> mattf, willb, mizdebsk, rsquared, pmackinn, samkottler  ^
15:04:12 <samkottler> hey hey
15:04:15 <mattf> hadoop is ftbfs
15:04:25 <mattf> in rawhide anyway
15:04:33 <pmackinn> nothing here
15:04:46 <willb> I'll be looking for a reviewer soon for something kind of ugly
15:05:10 <mizdebsk> hadoop has 2 issues afaik, 1) incorrect usage of %mvn_file macro and 2) incompatible version of netty
15:05:12 <rsquared> nothing from me
15:05:18 <tstclair> For me, the biggest issue is transitive jline dependencies.
15:05:27 <mizdebsk> 1) is a simple patch, 2) will probably require adding netty3 compat package
15:05:44 <rsquared> mizdebsk: A netty3 compat package is supposedly in the works
15:05:55 <pmackinn> netty36 compat
15:05:58 <rsquared> There's also jetty 9.1 issues with hadoop.  Possibly others
15:06:46 <rsquared> As of today there is a netty3 package in rawhide
15:07:28 <tstclair> rsquared, is the hadoop ftbfs due to jetty.   Are we going to carry a jetty8 compat?
15:08:07 <rsquared> tstclair: ftbfs is because mvn_file usage changed in rawhide, but that is just masking other other issues.  Any of which will produce ftbfs
15:08:27 <tstclair> willb, sbt review?
15:09:01 <willb> tstclair, indeed!  I think I've gotten it down to only two SRPMs for now, but it might need to go to three.
15:09:54 <tstclair> I don't think I would be the best reviewer on *this one, but are there any volunteers from the peanut gallery?
15:10:27 <tstclair> I think I recall samkottler interested in scala ^
15:10:44 <samkottler> yeah I can do some scala review work
15:11:04 <willb> samkottler, cool, we can discuss it offline; I'm thinking it will be up tomorrow sometime
15:11:17 <rsquared> samkottler: What's the status on the activemq update?
15:11:26 <samkottler> willb: cool I'm around all day so just ping and we can chat
15:11:33 <willb> thanks
15:11:47 <samkottler> rsquared: working on it still, hit a minor snag and got distracted with pressing upstream work for security stuff
15:11:48 <tstclair> mizdebsk, are there other jline transitive build issues besides bookkeeper?
15:12:12 <tstclair> I loath to update to jline2 b/c they drop it in their next version.
15:12:22 <tstclair> However I can't update due to HDFS issues.
15:12:26 <mizdebsk> i think there is a tracking bug for jline somewhere
15:12:42 <mizdebsk> but i don't remember the exact problems
15:13:58 <rsquared> tstclair: I believe carrying a jetty8 compat package is still an undecided.
15:14:32 <tstclair> rsquared, seems natural to me given the delta.
15:14:48 <rsquared> Thoughts from the BigData community would be welcome.
15:15:47 <mizdebsk> tstclair: i didn't found any other jline-related problems except bookkeeper ftbfs
15:16:01 <tstclair> :-(
15:16:08 <rsquared> imo, it's something we should decide soon.  Most of the ecosystem seems to depend on jetty in some form.  Changing now wouldn't be too bad, as it would mostly only hit hadoop.  hbase would need to change, but it's still in review.  The change for hadoop is non-trivial though
15:16:52 <tstclair> I'm a +1 for compat.
15:17:13 <tstclair> mizdebsk, do you have thoughts on the matter^?
15:17:30 <mizdebsk> i'm +1 for compat as long as you package only minimal set of modules (no systemd service et al)
15:18:09 <mizdebsk> the cost of porting to jetty 9 seems to outweight maintenance costs of jetty 8, and upstream doesn't want to migrate
15:18:23 <rsquared> mizdebsk: Forsee any issues with F20 hadoop using jetty9, then shifting to jetty8 in F21 or in an F20 update?
15:18:55 <rsquared> Upstream is definitely against jetty 9 because of the java 7 requirement.  jetty 8 seems a possibility though for upstream
15:19:22 <mizdebsk> idk about F20, i had F21+ in mind
15:19:50 <rsquared> As do I, but that still leaves the upgrade path for hadoop to be jetty9 -> jetty8
15:21:06 <mizdebsk> i don't see any problems here, that can be done even in F20, if needed
15:21:15 <mattf> so f20 hadoop deps on jetty (happens to be 9), f21 hadoop would dep on jetty8 -- different name and there's no problem?
15:21:48 <mattf> though i may be missing the mvn(...) non-version fun
15:21:52 <rsquared> And maintaining jetty9 in F20 would be difficult if F21 shifts to jetty 8.  The jetty changes are contained in a single integration patch which, in that past, wasn't easily able to be broken out to apply cleanly so we could patch for jetty9 in F20 and jetty8 otherwise
15:22:03 <mattf> but i'd imagine the compat would have to have a different name in the mvn dep too
15:22:37 <pmackinn> xmvn will resolve correctly with exact coordinates
15:23:19 <tstclair> #topic status_update
15:23:38 <tstclair> changing topics^ any news?
15:24:44 <mizdebsk> ivy local mode is in rawhide
15:25:00 <willb> mizdebsk, thanks!  I'll be bugging you soon for advice.
15:25:02 <mizdebsk> from now on you can easily build packages with ivy
15:25:04 <tstclair> mesos patches are going upstream, and I'm starting to work on issues.  Tachyon lead appears to be on semester break.. radio silence.
15:25:12 <rsquared> webapp deployment status: httpfs can be deployed and run with the tomcat@foo method, can not be deployed as webapp in system tomcat.  tomcat@foo and tomcat currently step on each other and there seems no co-ordination between the methods
15:26:48 <pmackinn> status - using ^^ said ivy changes for pig (then hive); assembly of prêt-à-porter docker images for BD developers to consume for their respective projects
15:28:26 <mattf> https://bugzilla.redhat.com/show_bug.cgi?id=1053464 - netty3 now available in rawhide thanks to  Marek Goldmann
15:30:50 <tstclair> #topic misc
15:30:56 <tstclair> anything else?
15:31:13 <tstclair> mattf, rsquared - anyone know what happened to the IBM folks?
15:31:21 <mattf> what would people think about having cascading?
15:31:33 <mattf> tstclair, good question, mia afaict
15:32:02 <willb> mattf, cascading seems useful as a thing in itself but is even more interesting for what it enables
15:33:38 <rsquared> tstclair: I haven't heard anything
15:33:41 <tstclair> Should we reach out to them, or just keep chugging.
15:34:40 <mattf> let's make an action to reach out
15:34:56 <tstclair> ack.
15:35:32 <mattf> nothing else from me
15:35:32 <tstclair> #action reach_out_to_ibm
15:35:40 <tstclair> zodbot fail.
15:36:08 <tstclair> anything else?
15:36:26 <rsquared> nothing from me
15:36:48 <tstclair> pmackinn, willb, mizdebsk ^ ?
15:36:53 <willb> nothing here
15:36:57 <pmackinn> nop
15:37:02 <mizdebsk> nothing from my side
15:37:15 <tstclair> ok, thanks everyone.
15:37:20 <tstclair> #endmeeting