14:00:56 #startmeeting freezer 14:00:57 Meeting started Thu Jul 7 14:00:56 2016 UTC and is due to finish in 60 minutes. The chair is m3m0. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:00:58 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 14:01:01 The meeting name has been set to 'freezer' 14:01:07 hello 14:01:13 Howdy. 14:01:16 hello, who's here for the freezer metting 14:01:19 hello all 14:01:22 * domhnallw is here 14:01:23 :) 14:02:05 o/ 14:02:17 daemontool you here? 14:02:37 He was in #openstack-freezer a few minutes ago... 14:02:41 o/ 14:03:23 o/ 14:04:15 ok let's start 14:04:26 #topic Improve the volumes backup through the Cinder APIs (i.e retention, delete, list backups, metadata) (domhnallw: pulled from last weeks' agenda) 14:04:33 who proposed this? 14:04:46 I think yangyapeng 14:04:48 ? 14:05:00 that totally make sense 14:05:10 hn 14:05:17 Yep, think so. I just did a copy pasta :) 14:05:52 we use freezer to manage cinder through Cinder APIs 14:06:18 cinder backup, delete, restore and so on . 14:06:23 ping iceyao_ 14:06:50 anyone is against improving the way we wrap the cinder API for volumes baas? 14:07:05 slashme, ping 14:07:23 No problem for me. 14:07:35 :) 14:07:43 yangyapeng, iceyao_ do you want to work on that? 14:07:48 yeah 14:07:50 The retention part would definilty be usefull 14:08:01 can you implement 1 API enpoint wrap at the time? 14:08:20 yeah we can. 14:08:22 I'm gald to 14:08:27 ok 14:08:51 can you add a basic delivery sequence in the freezer-specs repo 14:08:54 ? 14:09:32 like: 1) list backup, we to x y z, 2) metadata we save this xzv, 3) retention.... 14:09:34 ok? 14:09:43 very schematic 14:09:57 no paralysis anlysis if possible :) 14:10:29 sounds good? 14:10:57 ping ceyao_ 14:11:01 iceyao_: 14:11:21 you can take a commit next time. 14:12:15 ok 14:12:20 next? 14:12:31 unless anyone has anything on say about that 14:12:42 thanks 14:13:12 #topic Add support for file storage (Manila) (domhnallw: pulled from last weeks' agenda) 14:13:36 I added this last time 14:14:11 I think Manila soon will get way more relevance in OS 14:14:25 it would be one storage media more 14:15:26 is anyone interested? 14:15:35 ping yangyapeng 14:15:45 anyone thing it can be a priority? 14:16:20 I don't see it as a priority 14:16:22 ok 14:16:28 +1 14:16:29 Rather a nice-to-have feature 14:16:29 then disregard it 14:16:32 ok 14:16:35 now ,we need to do it manila ? 14:16:56 This is also something easier to integrate later when the agent refactoring is completed 14:17:00 ok 14:17:16 next 14:17:26 #topic Midcycle meetup 15 of July 14:17:43 next week is the mid cycle meetup 14:17:46 do we have an agenda? 14:18:43 link: https://etherpad.openstack.org/p/freezer_midcycle_meetup_newton 14:18:46 please review it 14:18:56 and add/remove topics accordingly 14:19:05 please let us know who propose changes 14:20:24 ok 14:20:58 can we move then? or should we discuss this ? 14:21:59 next 14:22:32 #topic Feedback from the Business 14:22:48 ok, I've added the last 2 topics 14:22:57 yesterday I had a very contructive conversation 14:23:14 with ATT 14:23:35 and basically they are evaluating 14:23:47 to use freezer or not in their production environments, allocating people etc 14:24:05 so they are concerned about the fact that we supports only tar 14:24:35 that also the deduplication was another point 14:25:09 the feedback was about also thinking of developing an additional component 14:25:14 for the storage 14:25:24 like the other enterprise solutions provide 14:25:45 to keep track of changed blocks, hash tables, etc etc 14:25:58 the positive feedback was about elasticsearch 14:26:29 it wasn't much welcome the idea of moving to mysql 14:26:47 the other poi of discussion was 14:27:07 about how to do backup, without moving the data form the back end storage, then process it and removing it to the storage again 14:27:13 as it is highly inefficient 14:27:32 one other point was the delay 14:27:52 of executions of jobs by the scheduler/agent 14:28:04 if we could find a way to reduce that delay to the minimum, it would be a big plus 14:28:10 also 14:28:21 they need a more advanced way for inclusion and exclusion 14:28:27 of files in the fs 14:28:32 when executing backups 14:28:41 +1 on the last point 14:29:00 honestly, I think all the points are valid 14:29:10 IMHO 14:29:32 thoughts? 14:29:39 what do you think? :) 14:29:40 I also agree. Some with higher priority though 14:29:54 DR wasn't covered 14:30:02 by there should be another conversation about it 14:30:09 sooner than later 14:30:31 adding engines should be easier than it is now 14:30:42 That's our goal 14:31:08 Ease the process of adding new engines, backends, snapshot mechanism 14:31:22 ah 14:31:32 As well as integrating with application (not only for backup but also for restore) 14:31:38 also the windows GUI was another point of discussion 14:31:59 as they most probably wants to try to reduce costs 14:32:07 not a strong point thought, but a nice to have 14:32:20 the strong point was about 14:32:21 engine 14:32:27 and backend&dedup 14:32:31 A Windows GUI? 14:32:34 yes 14:32:48 but let's not talk about that now please 14:32:52 it's more a business related thing 14:32:53 Okay. 14:32:55 we can talk about it offline 14:33:01 :) 14:33:16 so 14:33:18 at some point 14:33:25 as they are quite good writing C doe 14:33:26 code 14:33:30 there was the idea 14:33:48 of using a backend with zfs, that can export the changed blocks hashtable 14:33:58 then a component of freezer 14:34:00 on top of that 14:34:04 that interact with the agent 14:34:12 for dedup/incrementals 14:34:44 so these was basically the conversation 14:36:55 so, are they going to contribute soon? :) 14:37:13 they are evaluating 14:37:14 with us 14:37:22 netbackup, avamar 14:37:25 and someone more 14:37:28 can-t remember 14:37:35 If they see potential, they will 14:37:53 that could be great :) 14:38:06 I think that would make big difference 14:38:18 because their production environment 14:38:19 are huge 14:38:29 the requirements will be challening 14:38:50 but the solution will be way more advanced 14:38:52 also 14:38:55 in Ericsson 14:39:10 there's a test in the lab 14:39:15 to evaluate it 14:39:21 so expext that 2 engineers 14:39:24 will be around here 14:39:27 starting from next week 14:39:37 mostly working on docs 14:39:43 and tests 14:39:49 that is confirmed 14:39:55 Sounds good. 14:40:13 yes we need moe people 14:40:24 that's all from me 14:40:56 thanks a lot, we will help the new guys with everything they need 14:41:21 can we move forward? 14:41:39 yes 14:42:13 #topic Pending Reviews 14:42:19 https://review.openstack.org/#/c/329671/ (closes #1592046) 14:45:31 any more reviews? 14:46:40 can we close the meeting then? 14:47:12 I think so. Thank you m3m0. 14:47:18 Thanks. 14:47:25 thanks all :) 14:47:25 #endmeeting