14:00:05 #startmeeting sahara 14:00:06 Meeting started Thu Dec 21 14:00:05 2017 UTC and is due to finish in 60 minutes. The chair is tellesnobrega. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:00:07 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 14:00:10 The meeting name has been set to 'sahara' 14:00:14 o/ 14:00:48 o/ 14:02:40 waiting just a couple more minutes to see if anyone else shows up 14:04:36 I guess we can start 14:04:43 #topic News/Updates 14:05:05 I'm currently working mainly on image gen for ambari and mapr 14:05:09 most of time spend on downstream tasks this week. And also take time to solve problem in vanilla upgrading. hopefully can commit it in this weekend. 14:05:26 decommission is a little on the background until beginning of next year 14:05:47 that is good to hear shuyingya 14:06:15 I'm working on the new jobs, hopefully the native zuulv3 jobs for sahara-tests are ready (with one exception: the experimental multinode job, due to infrastructural limitations) 14:07:13 and I'm going to try to tackle the other missing jobs for sahara-extra, including publishing the oozie jar the cleanup of sahara-files (I'm tired from waiting for the changes in the publishing part, it can be fixed later) 14:08:05 sounds good 14:08:19 nice tosky 14:09:03 If I put the change, I will not commit SIE related changes util new jar can upload to tarball.openstack.org 14:09:19 I will try to have the job ready quickly 14:09:39 apart from that, I'm also catching up with a bug about CLI tests (thanks to Shi Yan for the ping) 14:11:36 if we don't have any more news, lets move on 14:12:01 #topic Priority List for next month 14:12:27 ... before Q-3 14:12:35 I talked with tosky yesterday about what we have to do for queens-3 14:13:47 we got down to we need to focus on image gen (i'm working on it), vanilla upgrade ( shuyingya ), APIv2, S3 14:14:11 am I forgetting anything tosky? 14:14:23 APIv2 because it's the biggest item on the list, and having it out at least as experimental would help 14:14:29 my colleague also focus on keypair injecting too. But he ask for a leave this week 14:14:46 cool 14:14:53 APIv2, of course absolutely not stable for this cycle, but at least we can start building on it and have it ready for Rocky (hopefully) 14:15:15 yeah 14:15:26 APIv2 is our highest priority for sure 14:15:34 S3 is a close second 14:15:54 the rest is up and close to finish 14:16:25 jeremy is working on those features, so I will get an update from him 14:16:38 but looks like it should be good as well 14:16:47 if we make all of those, it would be a good percentage of the plan from the PTG, which is not bad 14:17:01 yes 14:17:03 we can have then a big doc party before the release 14:17:04 we did plan big 14:17:12 wish we have a perfect release :) 14:18:05 btw, hbase integrated can be finished too before Q3 I think. 14:18:25 we can always learn from last cycle, we are going a great job, but I fear we aimed too high 14:19:19 but lets try to get the most of what we planned 14:19:49 and we have a direction for Rocky 14:19:52 thanks for tellesnobrega's good plan and leadship 14:19:58 true 14:20:09 shuyingya, I was just a part of it 14:20:14 especially removing APIv2 will help freeing resources (even if we would need to still stabilize it) 14:20:41 yes 14:22:35 I guess that closes the topic 14:23:00 #topic Open Discussion 14:23:35 do you know how to test a hive job? I didn't find it in sahara-tests' example 14:23:49 uhm, we may have had some issues with them 14:24:13 I want to ensure all of jobs can be successfully running on new vanilla 14:24:18 there is an old review about adding hive jobs for ambari which was consistently failing 14:24:45 this one: https://review.openstack.org/#/c/299358/ 14:24:58 check my comment 14:25:06 the one were I put -1 14:25:15 the review can be a starting point, I guess 14:25:36 thanks tosky 14:27:03 I also encounter problem that mapreduce job finished in Killed status 14:27:40 but the job was sucessfully finished in yarn and marked as success. I think it may cause by the swift interface issue. 14:28:07 not sure the reason in this patch, will investigate it 14:28:17 thanks shuyingya 14:29:42 Another thing is : do we have plan about ubuntu and centos or fedora image. I don't have too much energy to test all linux distribution 14:30:14 that's part of "rebuild the Sahara CI" 14:30:17 do you have some experience to share about the image test 14:30:47 oh, thanks tosky. At least, I will take effort to ensure sahara-ubuntu-image run 14:30:55 I would say priorities are ubuntu and centos 14:31:15 but try to get ubuntu done, and if you can't I will test out centos 14:32:03 yes, it is really hard to me to build an image domestically 14:33:22 I can help out for sure 14:33:35 thanks, tellesnobrega 14:35:05 when the jobs are converted, and multinode jobs work, we can automated at least the scenario tests for vanilla 14:35:43 right now the blocker is on the zuul+devstack side, and the incomplete support of multinode jobs 14:35:50 that should help a bit 14:36:03 that would be good enough for sahara 14:36:08 for a start 14:36:13 :) 14:36:34 once we have a third party CI we should add spark, cdh, mapr, ambari and storm 14:36:39 but one step at a time 14:36:42 yep 14:36:55 also, with zuulv3 we can reuse the same jobs 14:37:14 sounds good 14:38:47 ok, do we have anything else to discuss? 14:38:58 not from me 14:39:33 tosky? 14:41:04 tellesnobrega: nothing else from me 14:41:13 cool 14:41:28 thanks guys, we can call it short 14:41:56 we get 19 minutes back 14:42:05 #endmeeting