16:00:21 #startmeeting Cinder 16:00:22 Meeting started Wed Jul 10 16:00:21 2019 UTC and is due to finish in 60 minutes. The chair is jungleboyj. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:00:23 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 16:00:25 The meeting name has been set to 'cinder' 16:00:30 o/ 16:00:34 Agenda: 16:00:37 #link https://etherpad.openstack.org/p/cinder-train-meetings 16:00:38 hi 16:00:53 Courtesy ping: jungleboyj whoami-rajat rajinir lseki carloss pots woojay erlon geguileo eharney rosmaita enriquetaso e0ne smcginnis davidsha walshh_ xyang hemna _hemna 16:01:14 <_erlon_> Hey 16:01:14 hi 16:01:17 \o 16:01:19 hello 16:01:24 \o 16:01:26 @! 16:01:26 <_pewp_> jungleboyj \( ・_・) 16:01:27 hi 16:01:38 Hi 16:01:39 hi 16:02:58 Give people another minute to join and then get started. 16:03:03 Hey 16:04:52 Ok. I have enough coffee to make it through the meeting now. 16:04:56 So, lets get started. 16:05:04 #topic announcements 16:05:16 #link https://releases.openstack.org/train/schedule.html 16:05:31 Train milestone 2 is only about a week away. 16:05:46 A little over a week. 16:06:00 I think we made some progress looking at Specs last week. I will look through those again. 16:06:07 o/ 16:06:20 Do we have any drivers that look like they have a chance of making it? 16:07:25 Doesn't look like it. 16:07:56 jungleboyj, are you asking about new drivers, or the other deadline related to milestone 2? Because we are working on restarting the StorPool CI as we speak, it should be up and running in the next couple of days, with Python 3 jobs and all 16:07:57 o/ 16:08:22 So, don't really have any requests for driver reviews then since no one is updating the code. 16:08:57 Roamer`: Wasn't asking about Py37 requirement just yet but that is the next topic. 16:09:07 Glad you guys are making progress there. 16:09:20 There was some confusion as to the deadlines. 16:10:14 well, to be honest, we did notice that we have sort of forgotten to declare a couple of capabilities for the StorPool driver, and there is also an almost trivial way to implement snapshot revert... I *know* we may even be famous for coming in with last-minute request, but, supposing we manage to get our CI up and running and we submit some patches and it votes on them, is there a chance we 16:10:15 Your CI needs to be running py37 by Milestone 2 or an unsupported patch will be pushed up. 16:10:20 would make it? 16:10:45 If the py37 isn't passing by milestone 3 then the unsupported patch will be merged at that point. 16:11:25 Yes, I do believe last minute requests are a specialty. 16:11:36 Is there a new driver review out there to go with your CI? 16:12:41 Roamer`: ? 16:12:45 sorry, I'm not really sure exactly what you're asking right now... there is no new StorPool driver, there may be three very minor changes, we have not submitted them for review yet. 16:13:55 Oh, I thought you were asking about a new driver. 16:14:05 I see that storpool is still out there. 16:15:33 Ok, so it is out there and supported. 16:15:49 So, you just need to have your CI running using py37. 16:16:08 If you have fixes to push up then we can merge them once you are passing 3rd Party CI tests. 16:16:33 ok, thanks a lot, and sorry for blindsiding you yet again 16:16:39 Moving on. 16:16:55 I have gotten a Train release created in Launchpad. 16:17:10 Sorry I hadn't done that before. Had opened a page to do it several times and got distracted. 16:17:23 Anyway, it is there now. Take a look at it and let me know if there is anything that doesn't look right. 16:17:36 I need to update our PTL process documentation with a note on that. 16:17:47 #topic Mid-Cycle Meeting 16:17:58 So, we have had a bit of a hiccup in our mid-cycle plan. 16:18:39 We planned it for a date that worked for Sean and I to be in RTP to meet up with Locals, eharney hemna rosmaita tbarron 16:18:46 Was hoping to get others to come as well. 16:18:53 Now it sounds like smcginnis Can't make it. 16:19:12 Trying to figure out if we try to find another date or if we move on without smcginnis 16:19:54 eharney: hemna rosmaita Particularly looking for input from you guys. 16:20:11 Do you think we can have a productive meeting with 4 of us plus remote people? 16:21:07 *Hello? Is this thing on!?!* 16:21:17 yes, though it would be even better with smcginnis ... what are the date change options? 16:22:12 We could maybe move it a week later but then I think we lose hemna 16:22:32 I could try to do a week earlier but would need to move around some other vacation plans. 16:22:37 also, it's kind of late already 16:22:46 I am realizing that that date is coming up very quickly. 16:22:49 rosmaita: ++ 16:22:50 That date worked well for everyone else. I think you should just stick to the plan and I can try to participate remotely as much as I can. 16:22:54 will smcginnis be able to attend remotely at all? 16:23:00 oh, ok 16:23:08 smcginnis: You will be West Coast? 16:23:22 I'll be at an event that week, so I will have some distractions. But I can try to give my input inbetween things. 16:23:26 Yeah, CA. 16:23:53 Ok. So you could participate for a bit in the morning and we could try to schedule things that you need to give input upon around that? 16:24:08 #link https://etherpad.openstack.org/p/cinder-train-mid-cycle-planning 16:24:51 That would be fine with me. 16:24:53 hemna: eharney rosmaita Can you all confirm that you are still in? 16:25:04 yes 16:25:05 eharney: Were you able to get anyone else from Red Hat to join us? 16:25:09 yes 16:25:40 not likely, maybe jobernar 16:25:54 Ok. 16:26:16 So, if there is anyone else who can attend the meeting, please try to do so. 16:26:19 o/ 16:26:33 We will keep it on the planned date, 8/21, 22 and 23 . 16:26:37 No good OpenStack meetups in Raleigh this time? :) 16:27:25 We should start getting topics into the etherpad. 16:28:02 So, if you have things that you want to talk about please start adding them. 16:28:47 rosmaita: I will see if I can book the same hotel to make things more fun. 16:28:56 \o/ 16:28:58 Lenovo has deals with a lot of those hotels. 16:29:06 Page RD is a good area. 16:29:13 seemed reasonable price, no mention on vomit in the halls for this one 16:29:25 Hah 16:29:28 hemna: ^^^ See if you can maybe to do the same. 16:29:32 * jungleboyj laughs 16:29:42 We can change that. 16:30:15 BTW, I am hoping to have Zoom meetings by then which will help. 16:30:41 Ok, on to an even more contentious topic. 16:30:51 #topic Cinder PTG in Shanghai 16:31:13 So, the request for projects to state their interest in the Shanghai PTG has gone out. 16:31:28 It sounds like a very limited number of people are planning to go to SH. 16:31:40 #link https://etherpad.openstack.org/p/cinder-shanghai-ptg-planning 16:31:59 I have started our PTG planning Etherpad to get an indication who plans to be there more than anything. 16:32:06 Welcome to SH:) 16:32:29 LiangFang: is that where you are? 16:32:32 I am pretty sure I am not going to be able to avoid being there. 16:32:41 The ops meetup team recently did a Twitter poll to try to get feedback on number of attendees. I wonder if we should do something similar to try to get an idea of how many Cinder peeps will be there. 16:32:42 rosmaita: yes, I'm in Shanghai 16:32:51 smcginnis: You are planning on going. Correct? 16:33:10 I think so. 16:33:12 I could do that. Would bring public attention to this issue. 16:33:24 rosmaita: Is waiting to hear. 16:33:30 hemna: Is a no. 16:33:37 whoami-rajat: You are going to be there? 16:33:38 hopefully will know sometime next week 16:34:08 jungleboyj: hopefully, will be there. 16:34:20 Ok. So, that makes possibly 4 of us. 16:35:00 There are people that don't participate in this meeting who might want us to have the PTG though. 16:35:10 So the idea of a Twitter Poll isn't bad. 16:35:51 also, i specifically asked to be sent to the PTG ... don't know if they'll be willing to send me if we don't have it! 16:36:03 Anyone here think we shouldn't do a PTG slot? 16:36:29 i am pro-PTG 16:36:40 :-) Yeah, I think it would be bad for us to skip. 16:36:55 I think we would only need 2 days. 16:37:05 I would still do the Project update and Onboarding. 16:37:25 woojay: What did you ever hear about attending either of the events? :-) 16:37:25 Those are now part of the PTG I think I heard. 16:37:38 Right. If we have the PTG we can do those. 16:37:44 So, that is another reason to do it. 16:37:59 oh dang, I guess I better ask... 8-) probably PTG, I think. 16:37:59 from the ML, it sounds like they're expecting a lot of demand for onboarding 16:38:21 woojay: Ok, awesome. Update the etherpad when you have a chance. 16:38:33 rosmaita: That would be nice. 16:38:38 Sure. 16:38:42 Ok, so I am going to go forward with assuming that we will do it. 16:39:11 I will send out a poll and also a note to the mailing list and see what we get in the etherpad. 16:40:02 Anything else there? 16:40:42 Ok. Guess not. 16:40:49 So, last topic on the agenda 16:41:11 #topic Spec: Leverage compression hardware accelerator 16:41:13 LiangFang: 16:41:16 hi 16:41:35 Hi. 16:41:47 this is to leverage some hardware accelerator for image compression if such hw exist 16:42:00 Yep, remember discussing it at the PTG. 16:42:16 yes, Lisa carry this topic there 16:42:53 #link https://review.opendev.org/#/c/652275/ 16:43:03 I get some comment from eric and yangxin, and rosmaita 16:43:33 what should the format name be? gzip or gz? 16:44:05 Hmmm, thought I had read through this but guess not. I will do that today. 16:44:10 if it's always a gzipped raw image, .gz makes sense 16:44:27 eharney: ok 16:45:04 rosmaita said it should be a container format, not disk_format, i will take a look of it 16:45:55 hopefully my explanation on the review makes sense 16:46:28 LiangFang: Is there any clarification you need from rosmaita right now? 16:47:45 rosmaita, I will try to generate a spec-lite to glance, so the name of gz make sense to glance, right? 16:48:15 I feel like we had this discussion with Lisa at the PTG. 16:48:31 i think so ... would it make sense to have a 'compressed' format, and then the consumer figures out what kind of compression was used? 16:49:50 rosmaita: Might be good to be able to handle multiple compression types in the future. 16:49:59 do you mean not only support gzip, but also other format? 16:50:13 it may depend on how many different implementations of this people want to create 16:50:37 LiangFang: yes, we would only have to use gzip for your proposal, but the metadata in glance would say 'compressed' for the container_format 16:50:54 then glance won't have to add a different identifier for every format 16:51:00 That makes sense. 16:51:19 i believe most of them have a magic number in the header or something so you can tell what the compression format is? 16:51:20 ok 16:53:11 Anything else LiangFang 16:53:23 another thing is: should we fail it if there's no hw accelerator exist when user try to use a gzipped image? 16:53:40 no 16:53:42 LiangFang: No 16:53:58 i'd also like to point out https://review.opendev.org/#/c/668943/ , which is a different take on image compression, maybe of interest here 16:54:04 so we need to switch to software solution, right? 16:54:32 yes, it can just be ungzipped in software 16:54:52 eharney: ok 16:55:21 how about in the other direction? do we allow gzip on image create even if no hardware accelerator? 16:55:44 rosmaita: I would think so. 16:55:48 we should, there's no reason to tie this feature to a hard requirement on hardware 16:55:54 that was my thought, too 16:56:34 eharney: ++ 16:56:41 eharney: ++ 16:57:00 currently cpu is also very powerful 16:57:27 So, I think that covers that discussion. 16:57:35 Anything else we need to cover in the last 3 minutes? 16:58:04 should be add the tool name such as qzip and gzip to cinder.filter? 16:58:15 or let user to add by themself? 16:58:36 add to this file, so these tool can be run in root wrapped 16:58:55 Uhm, I think it probably be added to the filter so it can be rootwrapped. 16:59:11 ok 16:59:25 Ok. That is all the time we have for today. 16:59:30 is this assuming that the input is always a block device? 16:59:31 Thanks jungleboyj 16:59:31 thanks 17:00:05 Please add your topics to the mid-cycle etherpad and update your attendance for the PTG. 17:00:05 eharney: is it possible not a block device? 17:00:16 Thanks for joining the meeting! 17:00:18 yes 17:00:19 Should probably continue in the cinder channel. 17:00:25 smcginnis: ++ 17:00:30 #endmeeting