18:00:02 <vgridnev> #startmeeting sahara 18:00:07 <openstack> Meeting started Thu Aug 4 18:00:02 2016 UTC and is due to finish in 60 minutes. The chair is vgridnev. Information about MeetBot at http://wiki.debian.org/MeetBot. 18:00:08 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 18:00:11 <openstack> The meeting name has been set to 'sahara' 18:00:33 <vgridnev> #link https://wiki.openstack.org/wiki/Meetings/SaharaAgenda 18:00:52 <vgridnev> hello folks! 18:00:55 <elmiko> hi 18:01:04 <egafford> o/ 18:01:08 * tosky here but no here 18:01:33 <tellesnobrega> o/ 18:02:53 <vgridnev> tosky, as a well known schrodinger's cat 18:04:06 <vgridnev> #topic News / updates 18:04:16 <elmiko> lol 18:04:25 <elmiko> nothing new to report from me 18:05:00 <vgridnev> folks, let's vote for summit talks, voting is until Aug 8 18:05:04 <egafford> Documented the image gen CLI; working on (WIP at first) patch sets to generate images for Ambari, Cloudera, and MapR using it. 18:05:25 <elmiko> vgridnev++ 18:05:48 <tellesnobrega> I still haven't got many reviews on the floating ips refactoring patch. egafford's one I already responded to them, but it would be nice to get some eyes there 18:06:39 <vgridnev> egafford, would love to see changes 18:06:51 <egafford> vgridnev: Me Too. 18:07:27 <vgridnev> I'm actually working on Kerberos impl, base stuff (deploying infra is on review, cloudera support is on the way) 18:07:29 <elmiko> sorry tellesnobrega, i didn't make it to that one. i did ping tmckay about it though 18:07:46 <tellesnobrega> elmiko, awesome, no rush 18:09:04 <vgridnev> can't find NikitaKonovalov at the meeting (probably he is celebrating birthday) but he was working on boot from volume support (spec is on review) and some internal stuff 18:09:31 <elmiko> happy birthday NikitaKonovalov ! 18:09:58 <tellesnobrega> happy bday to him, i will try to take a look on the spec today 18:11:28 <vgridnev> #topic Boot from Volume 18:11:31 <vgridnev> #link https://review.openstack.org/#/c/349516/ 18:11:44 <vgridnev> let's review this spec, it's quite short actually 18:12:17 <elmiko> ack, looking 18:15:51 <elmiko> that seems reasonable to me 18:16:09 <tellesnobrega> to me too 18:16:18 <egafford> +1 18:18:12 <vgridnev> that is cool. will approve that after the meeting then 18:18:13 <egafford> I think we're done, actually. That was indeed quick. 18:18:17 <egafford> :) 18:19:35 <elmiko> brb 18:19:59 <vgridnev> elmiko, do we need to api v2, just interesting? I think that we can move that to backlog until the summit 18:20:51 <elmiko> yeah, backlog sounds appropriate 18:21:01 <vgridnev> ok 18:21:10 <vgridnev> #topic Open discussion 18:21:12 <elmiko> sorry, i wish i had more time for it 18:22:23 <tosky> if you noticed, I'm investigating the status of embedded jats 18:22:36 <tosky> jars* 18:22:36 <vgridnev> elmiko, I think that we need to collect everything about that, describe some tasks, and move that for outreachy team 18:22:40 <vgridnev> is it possible? 18:23:01 <vgridnev> tosky, I noticed 18:23:06 <elmiko> definitely possible, the wiki page and spec describe all the work 18:23:26 <tosky> there is an old bug from debian, and also RDO packagers requwsted details 18:23:44 <tosky> so expect some questions :( 18:23:49 <tosky> :) 18:24:06 <tosky> like "where did it get this from" 18:24:10 <tosky> eom 18:24:18 <egafford> tosky: Is it evil in packaging-land to have a .jar with the source embedded in it? 18:24:18 <vgridnev> elmiko, do you know some contacts to reach that? probably if we will write something to ML we will get some response about that 18:24:44 <egafford> Our .jars only have the .class files, but you can easily build .jars with .java source as well. 18:25:06 <egafford> (I figure it's probably still evil and we should just build from source like real people, but.) 18:25:10 <elmiko> vgridnev: contacts for outreachy? 18:25:18 <tosky> it is, it should all buildable from sources 18:25:23 <vgridnev> elmiko, yes 18:25:56 <tosky> anyway, I will document them to start 18:26:15 <vgridnev> tosky, wrapper #1 https://github.com/openstack/sahara-extra/tree/master/edp-adapt-for-spark 18:26:25 <elmiko> vgridnev: 1sec 18:26:25 <tosky> that was easy 18:26:35 <vgridnev> tosky, wrapper #2 https://github.com/openstack/sahara-extra/tree/master/edp-adapt-for-oozie 18:26:50 <tosky> I was hunting the source for upstran hadoop examples 18:26:51 <elmiko> vgridnev: i would start here https://gnome.org/outreachy/ 18:27:00 <elmiko> and then https://mail.gnome.org/mailman/listinfo/outreachy-list 18:27:15 <elmiko> also, vkmc was my main contact for working with the outreachy group 18:27:33 <egafford> tosky: Never say "that was easy" if someone points you to a repo with a pom.xml file. :) 18:27:55 <egafford> (Though these are really simple Maven projects, admittedly.) 18:28:38 <elmiko> vgridnev: one more link, https://wiki.openstack.org/wiki/Outreachy 18:30:21 <vgridnev> find source of truth for jobs is much harder task 18:34:14 <egafford> elmiko: Yeah, I should ping vkmc about getting involved in Outreachy. 18:34:54 <egafford> It'd be good to bring more interns in, for suure. 18:35:15 <elmiko> definitely, she is absolutely top notch 18:38:38 <vgridnev> tosky, do you think that we have to put source of jars in tree of sahara? 18:38:43 <vgridnev> and sahara-tests 18:39:39 <vgridnev> there are probably some issues with that for sahara-tests, we are using a udf jar (user defined functions for pig), I think that is not so good 18:40:41 <egafford> vgridnev: For packaging purposes, we'd probably want to package the java packages separately and make them dependencies (the specs would be quite simple.) 18:40:59 <elmiko> i have to step away 18:41:23 <tosky> we have some of the sources, we should have all 18:41:46 <tosky> and an easy way to build then 18:43:30 <egafford> tosky: Are you imagining we'd want to embed the java source in the Sahara repo? 18:44:07 <tosky> or sahara-extra maybe it's enough; not sure yet 18:44:47 <vgridnev> I think that we can put that to sahara extra, and prepare script for building 18:45:29 <egafford> Agreed; it feels less awful to embed Java trees in sahara-extra than sahara. 18:45:34 <vgridnev> that is possible, except for hadoop-mapreduce-examples-2.6.0.jar and udf.jar for sahara-tests, I think 18:47:04 <vgridnev> we also can create a publisher which will be building that and will collect that at tarballs.openstack.org 18:48:29 <tosky> could be a good idea 18:49:14 <vgridnev> so, let's create a final plan. 18:49:34 <vgridnev> collect code from jars as much as possible at sahara-extra 18:49:52 <vgridnev> setup building scripts for those 18:50:11 <vgridnev> setup publishers 18:51:14 <tosky> good :) 18:51:45 <egafford> Sounds like the plan. 18:51:46 <vgridnev> create a config option for sahara for placing jars files (so that sahara will know to get those) 18:52:03 <egafford> Packagers can go straight to the build scripts from sahara-extra. 18:53:32 <vgridnev> or we can specify a url where to get them 18:54:57 <vgridnev> for hadoop-mapreduce-examples-X.jar I think we can consume a jar from instances (I will be on all installations, sahara-tests can do so) 18:55:18 <vgridnev> (I will be -> it will be) 18:56:10 <tosky> if there is on the instance, yes, use that 18:56:16 <vgridnev> we can put that to hdfs to some location and then pass that as a job binary 18:56:31 <egafford> vgridnev on all installations would be a significant feature. :p 18:56:45 <vgridnev> :) 18:57:48 <egafford> vgridnev: But yeah, for the mapreduce-eamples I think it's ok to use a URL in the main case (since we're consuming the publication artifacts of another OSS process, and we don't need it in the production workflow case.) 18:58:56 <vgridnev> egafford, yep. we can try to search that on instance, and if failed, get that from url 18:59:14 * egafford could be wrong, but maintaining our own effective fork of the hadoop examples seems totally unnecessary. Yup, that fallthrough makes sense. 18:59:29 <egafford> Give people an extension point (one they'll probably never use, but.) 19:00:14 <tosky> also for swiftfs, it would be nice to remerge, but that's a more complicated story 19:00:15 <vgridnev> ok, we are out of time, thanks folks 19:00:20 <egafford> o/ 19:00:27 <vgridnev> #endmeeting