14:03:20 #startmeeting sahara 14:03:21 Meeting started Thu Mar 29 14:03:20 2018 UTC and is due to finish in 60 minutes. The chair is tellesnobrega. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:03:22 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 14:03:24 The meeting name has been set to 'sahara' 14:03:41 o/ 14:04:02 o/ 14:05:35 sorry I was late 14:05:37 :) 14:05:40 lets start 14:05:51 i have to leave at 45 past the hour, btw 14:06:01 should be enough time 14:06:11 #topic News/Updates 14:07:20 I'm working on plugins split, facing some problems working on the API 14:07:36 but it is better, and we should have a working prototype with api soon 14:07:40 while the review which fixes the multinode/spark job is ready (reminder: https://review.openstack.org/#/c/547669/, on par with the current legacy job), 14:07:53 I'm working on a multinode job (vanilla), the idea is to have some periodic jobs 14:08:10 still been pretty busy with non-sahara things, but i do have some small patches in the works (beginning of admin-manage-all, some bugfix inspired by downstream)... more interestingly, i've finally acquired some hardware for our external CI (this will be a new meeting topic) 14:08:12 and it may be reused in a future CI (with zuul integration and so - zuul v3 is out officially!) 14:08:30 (see https://review.openstack.org/#/c/556846/ ) 14:09:54 thanks for the work on spark job and vanilla tosky 14:10:04 jeremyfreudberg, cool, let me know if you need anything 14:10:17 I guess we can move on 14:10:30 yup 14:10:45 #topic Sahara third-party CI \o/ 14:10:58 the title says it all 14:11:37 right, I managed to receive a "donation" from Cisco 14:11:48 yesterday we got an email with some pretty good news with some machines for our CI 14:12:23 jeremyfreudberg, should be our contact on getting them ready for our CI 14:12:37 two nodes with Xeon E-2630 and one with 192 gb ram, the other with 160 gb 14:13:05 are those machines connected together, so that they can be used as small cluster? 14:13:14 should be enough memory for our case 14:13:56 tosky, yes, that's the idea, i'm installing them both in the right rack later today 14:14:23 (i will point out that these nodes are a bit heavy on ram and slightly light on cpu, but it should be fine) 14:14:35 the hardware will handle itself here, we need to talk about software though 14:14:55 do we have some limitations/requirement on one side, and ideas on the implementation on the other? 14:15:25 about the latter point, we had sahara-ci-config, but I'm not sure about them and zuulv3 14:16:01 so, we do have sahara-ci-config, there's a few things in there hardcoded for mirantis stuff 14:16:16 and i don't think much progress was made toward zuulv3 there 14:16:25 jeremyfreudberg, do you know how complicated would be to fix it up? 14:17:01 I guess my first question is, do we want/need zuulv3? 14:18:31 my answer would be: if not, what else? 14:19:01 well legacy zuul still works, and it would be nice to have something up and running faster 14:19:56 how different is to deploy zuul and zuulv3? 14:20:09 no idea 14:21:36 tosky, any idea? 14:21:39 iirc software factory already supports zuulv3 14:21:47 even more now that it's released (earlier today) 14:22:05 https://blogs.rdoproject.org/2017/11/getting-started-with-software-factory-and-zuul3/ 14:22:16 my vote would be to go with zuulv3 14:23:23 we won't need to upgrade any time soon 14:24:53 i think having zuulv3 is important, but certainly it should come last (have everything else working first) 14:25:21 right, but is the effort to deploy with the current script less than setting with zuul v3? 14:25:28 if yes, maybe, ok, zuul v2.x 14:26:00 true 14:26:38 i'm really not sure how the scripts work, and which components could be superseded by switching to v3 14:28:01 ok we can go back and forth on this forever, let's try to make a concrete list of steps 14:28:04 SotK, I would say the best course of action here is, take some time to study sahara-ci-config and figure out the best way to deploy sahara-ci 14:28:11 again with the ping 14:28:25 I hate that 14:28:26 sorry 14:29:09 0) jeremy sets up the machines with OS, public network, etc 14:29:11 1) ...? 14:29:35 :) np 14:29:37 1) figure out ASAP the best way to deploy sahara-ci 14:30:26 1a) sahara-ci in terms of components (zuul, jjb, devstack) 14:30:52 1b) sahara-ci in terms of topology (can we have master and slave on same machine to increase capacity) 14:31:31 I agree that zuul should not take all the resources 14:31:39 +1 14:33:01 also I don't expect to see tons of jobs as on the main CI, so it should not require too much resources 14:33:48 about components, for reference: the old CI had a fixed devstack instance which was refreshed every month or so, and only sahara was redeployed on the fly 14:34:02 which makes total sense 14:34:16 that is basically what I have at home 14:34:27 yes, the persistent devstack i think is something important (and non-standard) 14:34:35 I have a devstack running and I connect my local sahara to it 14:34:40 which is faster, but I'm not sure that allows co-gating 14:34:58 not co-gating, sorry: I mean the usage of Depends-On 14:35:37 well, it can depend on other sahara patches only i think 14:35:57 but again, i'm not exactly sure 14:36:11 we do need to talk to evgeny who knew some of the intricacies 14:36:20 that is probably best 14:36:36 jeremyfreudberg, can you email evgeny so we have this all on ML? 14:37:19 yes, i'll send something to ml/evgeny 14:37:33 with both the general announcement and the questions/planning bits 14:37:45 #action jeremyfreudberg to formally email evgeny about Sahara CI 14:37:52 do we have other topics for today? 14:38:06 I don't 14:39:09 just news from shuyingya, he is not around because he's got a fever and is out of work today 14:39:38 sorry to hear that (feel better, shuyingya, if you are reading the logs :D) 14:39:46 tosky, any topics you want to cover today? 14:41:07 just a reminder for shuyingya if he reads this: I added a minor comment to his patch, then it could go in 14:41:24 and I'd like to point out that we had a new contributor sending a patch for HDP 2.6 14:41:28 we got a patch with HDP 2.6 14:41:33 you beat me to it 14:41:45 it looks good btw 14:41:48 i think that contributor might be gokhan 14:41:49 (which probably requires a new patch for sahara-image-{elements,pack}) 14:41:58 or a colegague of gokhan 14:41:59 or gokhan's team 14:42:01 yup 14:42:07 yup to tosky also 14:42:16 but it was not the only team interested in ambari 2.6 14:42:27 yes, but the other was russian, this is turkish 14:43:16 oh, right 14:43:51 nice, bottom line is, we got new contributors which is great :) 14:44:03 yes, i'm very happy about that 14:44:07 #topic Open Discussion 14:44:19 did any of us ever tested castellan? 14:44:33 ehm, we have an open bug 14:44:40 castellan is on my list for this cycle, i haven't tested it since we discussed with michael a few months ago 14:45:02 ok, I figured on plugins, there is key_manager.delete_key(key_id, ctx) 14:45:06 that is from castellan 14:45:31 but, the correct method name is delete_secret(...) 14:45:40 hmm, i will have to look into that 14:45:55 probably never worked 14:46:17 #action placeholder action to remind jeremy about more recent castellan bitrot discovery 14:46:31 gtg, thanks all 14:46:36 thanks jeremyfreudberg 14:46:43 we can close the meeting 14:46:49 thanks tosky 14:46:51 ok, about time - there will be few holidays around in the next days, so happy holidays for people who have them! 14:47:00 happy holidays 14:48:53 o/ 14:49:02 #endmeeting