16:01:18 #startmeeting cinder 16:01:21 Meeting started Wed Jan 31 16:01:18 2018 UTC and is due to finish in 60 minutes. The chair is jungleboyj. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:01:22 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 16:01:24 The meeting name has been set to 'cinder' 16:01:26 Hallo 16:01:28 o/ hello 16:01:32 o/ 16:01:36 o/ 16:01:36 Courtesy ping: jungleboyj DuncanT diablo_rojo, diablo_rojo_phon, rajinir tbarron xyang xyang1 e0ne gouthamr thingee erlon tpsilva patrickeast tommylikehu eharney geguileo smcginnis lhx_ lhx__ aspiers jgriffith moshele hwalsh felipemonteiro 16:01:38 o/ 16:01:41 @! 16:01:41 <_pewp_> jungleboyj (♦亝д 亝)ノ 16:01:41 Hello 16:01:45 hey 16:01:45 hi 16:01:46 hi 16:02:05 smcginnis: Sorry, I am having an issue in IRC cloud where it lags when trying to change channels. 16:02:10 :) 16:02:13 o/ 16:02:20 Keep hoping they will fix it but it eats up a crap load of processor, whatever it is doing. 16:02:37 * jungleboyj continues to be addicted to IRCCloud anyway. 16:03:14 jungleboyj: works well on my Mac ;) 16:03:53 ildikov: Eeeew. You said Mac. 16:03:57 Bourgeoisie. I just use hexchat. 16:04:07 hexchat FTW 16:04:09 hey 16:04:11 Swanson: The minimalist 16:04:22 irssi FTW! :) 16:04:28 ildikov: everything works on mac except cinder's unit tests:( 16:04:30 smcginnis: The massive nerd 16:05:09 Ok, we should get started. 16:05:11 e0ne: not sure only the Mac is the issue in that case :) 16:05:20 e0ne: but I hear ya 16:05:20 ildikov: Ouch. 16:05:21 ildikov: :) 16:05:28 #announcements 16:05:35 #topic announcements 16:05:49 * jungleboyj blames the super uber massive blue moon thing. 16:06:01 Q-3 was cut last week. 16:06:11 We are now in a short period of time to merge bugs. 16:06:36 Nicely defined unit of time. 16:07:28 :-) 16:07:39 So, please make sure to not be merging any new features. 16:07:48 Please also work on merging bugs. 16:08:05 #link https://releases.openstack.org/queens/schedule.html 16:08:12 RC-1 is next week. 16:08:26 I am guessing that means that I need to cut RC-1 next THURSDAY? 16:08:29 smcginnis: ??? 16:08:41 jungleboyj: Yep, Thuuuursday. 16:08:53 :) 16:08:55 Would that be the day after Wednesday? 16:09:00 :) 16:09:01 :) 16:09:11 See, I can be taught. 16:09:13 The day before Friday. 16:09:17 My coal fired jebus that is coming up quick. 16:09:20 I love Fridays. 16:09:25 Swanson: Yeah. 16:09:41 jungleboyj: usually, Thursday is between Wed and Fri, but who knows... 16:09:46 :) 16:09:53 e0ne: I know nothing anymore. 16:10:05 PTL Nominations are now open. 16:10:14 #link http://lists.openstack.org/pipermail/openstack-dev/2018-January/126700.html 16:10:35 That has also come up quick. 16:10:50 And finally a reminder about the PTG planning etherpad: 16:10:59 #link https://etherpad.openstack.org/p/cinder-ptg-rocky 16:11:09 Looks like we have a good crowd coming and we are getting topics. 16:11:11 Yay! 16:11:20 What ever happened with longer cycles? 16:11:40 FYI, ticket prices raise on Thursday 16:11:43 I am also planning to go back to the notes from the Queens PTG and pull forward things that didn't get addressed during the release. 16:11:43 Swanson: TC decided to not change anything at least for Rocky 16:11:45 Swanson: Too many issues were raised with extending it. 16:11:50 ildikov: ++ Thank you! 16:12:38 Any other announcements? 16:13:16 Ok, moving on. 16:13:26 #topic Feature Checkpoint 16:13:43 Transfer Snapshots with Volumes was too big a change. 16:13:52 Anyone strongly object to not merging it? 16:14:23 I will take that as a no. 16:14:51 Thank you for the work to get Provisioning and RBD encryption changes in. 16:14:54 Those merged. 16:15:09 Looks like there will be a spec for the RBD change here later today. 16:15:10 Thank you. 16:15:18 eharney: Just ping me when that is up. 16:15:28 yep, will do 16:15:32 :-) 16:15:48 #topic Project Interviews at PTG. 16:15:58 Taco run? 16:16:02 So, they are interviewing people from the project again. 16:16:09 https://docs.google.com/spreadsheets/d/1MK7rCgYXCQZP1AgQ0RUiuc-cEXIzW5RuRzz5BWhV4nQ/edit#gid=0 16:16:22 Is there anyone willing to be my backup incase I get stuck on a Taco Run? 16:16:27 :) 16:16:34 smcginnis: You had to beat me to the joke, didn't you? 16:16:44 I nominate ildikov. 16:17:03 smcginnis: :-) She can talk about how wonderful it is that we got Multi-attach in. 16:17:05 Do a dance! 16:17:31 * jungleboyj is also going to figure out where the heck the interviews are happening ahead of time this time. 16:18:50 So, I guess I will arrange a time for Cinder. Not near a meal time and see who wants to join me when the time comes. 16:19:02 I do think that ildikov is a good choice though. :-) 16:19:15 Even though she is ignoring us. 16:19:49 I didn't hear an objection from her, so I think that means you can sign her up for it. ;) 16:20:03 smcginnis: I like the way you think when you aren't harassing me. 16:20:37 Hah 16:20:39 jungleboyj: I only do it with my dream team jgriffith and patrickeast :) 16:20:57 ildikov: Ouch. You cut me deep. 16:21:25 jungleboyj: I think we did an awesome job last time :) 16:21:51 Niether of who are on the attendance list though. 16:21:52 jungleboyj: you need to learn to delegate ;) 16:22:08 jungleboyj: yeah, you need to fix that! 16:22:09 ildikov: Ok, I will sign you guys up then. 16:22:17 jgriffith: Where you at!?! 16:22:53 I will sign us up and we can figure out the details. 16:23:24 #topic Cisco FC ZM Driver 16:24:15 So, Cisco is working on getting CI working. 16:24:29 It isn't pretty but it is reporting and they have been working on this. 16:24:31 They do have one (and only one) test running. 16:24:44 smcginnis: Ok, it is something. 16:25:02 I wanted to get an idea of what you think smcginnis and if anyone else has opinions. 16:25:27 I had worked with Angela in the past and we came up with a list of 13 tests that actually go through the fczm. 16:25:37 I think that should apply to Cisco as well. 16:25:48 I'm ok with this if CI is stable and reporting to the needed patches 16:25:49 I need to try to find that though. That was like 2 years ago. 16:26:08 e0ne: OK with it now with one test? 16:26:20 Not disagreeing, just clarifying. 16:26:33 smcginnis: That sounds reasonable. They should do the same thing as the other FC ZM. 16:27:12 That said, I don't think Brocade is reporting anymore now either since the Broadcom acquisition. 16:27:26 smcginnis: *Laughing* 16:27:29 smcginnis: one tests doesn't look good enough 16:27:56 I would be willing to accept "good enough for now" with the understanding that they need to add more tests. 16:28:13 I'm not expert in FC at all, but I would like to see FC-related tests with some volume backend tests 16:28:17 smcginnis: ++ and then we should follow up with Angela on Brocade. 16:28:48 If either is still around... 16:29:02 * jungleboyj shakes my head 16:29:15 e0ne: At least the one they have running is volume attach/detach, so it is at least covering the main happy path. 16:29:24 smcginnis: I feel like marking it unsupported had the desired effect. 16:30:01 True 16:30:25 smcginnis: Do you have bandwidth to continue working with Liqin Dong? 16:30:45 Thermo nuclear option, we remove FCZM from cinder and it's up to Cisco and Brocade/Broadcom to provide this functionality to their end users. I don't like that option though. 16:31:21 I am not a fan of Thermo Nuclear options. 16:31:28 jungleboyj: Yeah, I'll try to track down that conversation with Angela and get them the list of expected tests. 16:31:42 ACtually, now that I think of it, I might have been smart enough to add that to the wiki. 16:31:45 * smcginnis checks... 16:31:59 smcginnis, I like this. Let's give it the ol Kim Jong Un. I can have a patch up for my driver in 15 minutes. 16:32:32 smcginnis: attach/detach - is good to start. from my experience, most problems were with FC and migrations. but we didn't use FC zone manager in that deployments 16:32:32 * jungleboyj shakes my head at Swanson 16:33:03 e0ne: Oh, good point. 16:33:14 #action smcginnis to follow up with Cisco on tests being run by Broacde. 16:33:28 #action jungleboyj To follow up on state of the Brocade driver. 16:34:18 That all sound good? 16:34:23 ++ 16:34:29 Sweet. 16:34:41 #topic Provisioning report bugs: 16:34:47 tpsilva: Take it away. 16:34:52 alright 16:35:22 so, as part of the provisioning improvements, we went through all the drivers checking what they were reporting 16:35:23 Alright ... 16:35:57 and we came up with a list of bugs 16:36:19 for each driver that had incorrect provisioning reports 16:36:34 most of the bugs are pretty simple to fix, so it shouldn't take much 16:36:37 tpsilva: Excellent. 16:36:48 but vendors should check this list to see if their drivers are affected 16:37:24 #action Vendors need to check if their drivers are impacted and make changes. 16:37:41 it sounds like we could just push these changes rather than requiring driver owners to make them, if the fix is to just not report a value from the driver anymore 16:37:50 and we should be careful with RemoteFS driver since it can broke some other drivers 16:38:09 eharney: I am ok with that. 16:38:30 tpsilva: I think I saw about 5-6 bugs go through. Does that sound right? 16:38:45 smcginnis: I opened 8 bugs 16:38:51 eharney, some of the changes deal with some internal driver logic and we didnt wanted to mess with that 16:38:55 the complete list is on the meeting agenda etherpad 16:39:04 tpsilva: OK, thank you. 16:39:07 smcginnis: List is in the meeting notes. 16:40:54 eharney, and, changing drivers without having the hardware to test means to rely on their CIs, and that use to be a pain 16:41:19 erlon: What is the impact of leaving them as is? 16:41:20 erlon: well that is one of the main reasons that we have all of these CIs... 16:42:22 eharney: True. Assuming they work. 16:42:51 don't they? 16:43:01 eharney: Some do. :-) 16:43:13 eharney: That is a whole other discussion though. 16:43:17 jungleboyj, true 16:43:55 seems like a reasonable assumption given the numerous man-years put into dealing with them, but i dunno 16:43:58 erlon: Did you see the question about what the impact of leaving drivers as they are? 16:44:08 jungleboyj, I have done some work through several drivers and waiting on CIs to get things tested is very time consuming 16:45:06 eharney, the impact is that those drivers will not behave correctly when using thin provisioning, some will not schedule when they should, other will always create volumes etc 16:45:25 erlon: Ok. 16:45:50 the idea of have then fixed is to have a consistent behaviour on thin provisioning supported drivers 16:47:18 erlon: So, I am looking at this list and it shouldn't be hard for those drivers to get people involved to make sure the right change is pulled in. 16:47:36 Huawei is one of them. I know someone with connections there. :-) 16:47:43 Storwize, same thing. 16:48:01 So, let me take this and follow up with people to get it attention. 16:48:31 Keep this on the meeting agenda to track progress. 16:48:42 I don't think we want drivers with inconsistent behavior. 16:48:47 jungleboyj, nice! Its good to have vendors involved 16:49:17 erlon: Yeah, I feel they should be involved. If you have patches, that is great, they should review them. 16:49:37 Lucian is active on SMBFS. 16:49:55 eharney: e0ne You guys have the most experience with remotefs. Right? 16:50:49 yes? 16:52:07 That is the one kind-of concerning one as it impacts multiple drivers. 16:52:15 erlon: How complicated is the change there? 16:52:49 jungleboyj: I can try to fix remotefs 16:52:58 e0ne: Excellent. 16:53:02 jungleboyj: we just need to remove the report for allocated_capacity_gb 16:53:09 not that complicated 16:53:13 tpsilva: Ok, good. 16:53:23 piece of cake :) 16:53:26 So here is what I have: 16:53:32 but any input from tpsilva will be extremely useful 16:53:47 #action jungleboyj to make vendors aware of bugs against their drivers . 16:54:06 #action erlon to push up any patches he has started and to look from input from the vendors 16:54:22 #action e0ne to look at the remotefs driver. 16:54:47 I assume all these changes are things that could be backported in the case they don't go in before RC? 16:54:54 jungleboyj, Ill send an email for the maintainers, dont have any patches tough 16:55:10 erlon: Oh, you are going to e-mail the vendors? 16:55:46 erlon: That would be awesome if you can do it. 16:55:57 jungleboyj, wasnt that what you meant by 'make vendors aware'? 16:56:23 need to head home. I'll be online in ~1.5 hours 16:56:36 erlon: Yep, that was it. If you are able to do it, that is great. 16:56:51 jungleboyj, ok, I wll 16:57:02 erlon: Excellent. Thank you! 16:57:27 #action erlon to coordinate with driver maintainers to make sure that drivers are properly fixed. 16:57:34 erlon: Let me know if you need help. 16:57:38 I had one more topic. 16:58:06 #topic Bug: Volume details shows attached compute host for non-admins 16:58:09 jungleboyj, np 16:58:22 #link https://bugs.launchpad.net/cinder/+bug/1740950 16:58:23 Launchpad bug 1740950 in Cinder "Volume details shows attached compute host for non-admins" [Undecided,New] 16:58:36 This bug was opened a bit ago and could be a security issue. 16:58:45 Doesn't look like anyone has had a chance to look at it. 16:58:52 Don't think we want to release it as is. 16:58:59 Anyone able to investigate it? 16:59:31 * jungleboyj hears crickets 16:59:39 Just wanted to point, while we are feature and string freeze, now would be a good time to spend time reviewing pike backports. That transitions to Phase II support in just a few weeks. 16:59:51 jungleboyj: I can try to take a look later unless someone else beats me to it. 17:00:02 Which would be just fine with me. 17:00:07 Times up. 17:00:18 smcginnis: Thank you and thanks for the note on backports! 17:00:40 Thanks everyone! Talk to you all next week! 17:00:44 #endmeeting