15:00:11 #startmeeting manila 15:00:12 Meeting started Thu Feb 5 15:00:11 2015 UTC and is due to finish in 60 minutes. The chair is bswartz. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:00:13 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:00:16 The meeting name has been set to 'manila' 15:00:23 hello all 15:00:32 hello 15:00:38 hi 15:00:40 hi 15:00:48 hi 15:00:49 #agenda https://wiki.openstack.org/wiki/Manila/Meetings 15:01:09 hi 15:01:31 ok 15:01:39 some of you may have seen my ML post 15:01:45 K-2 was completed last night 15:01:48 hi 15:02:11 thanks again to those who helped out with reviews 15:02:14 \o 15:02:21 #topic kilo-3 15:02:42 so now stuff targeted for K-3 can start merging 15:02:44 hi 15:02:59 you can read my ML post for reminders about the deadlines 15:03:16 #link https://launchpad.net/manila/+milestone/kilo-3 15:03:33 there's a lot of stuff here 15:03:45 this list needs to get cleaned up 15:04:11 really there's only 4 weeks left to wrap up features so I'm sure some of the things there simply won't make it 15:04:36 do yourself a favor and submit early so your feature isn't one that doesn't make it 15:05:04 after Kilo-3, there will be no new features until L 15:05:38 okay enough on that 15:05:41 #topic dev status 15:05:51 dev status: 15:05:55 1) New standalone network plugin 15:05:59 BP: #link https://blueprints.launchpad.net/manila/+spec/standalone-network-plugin 15:06:03 gerrit: #link https://review.openstack.org/151640 15:06:04 status: ready for review 15:06:11 2) Generic driver now has two driver modes - with and without share servers handling. 15:06:18 3) Manila has functionality "level of access for shares". Some new drivers have it, but all existing should start using it where possible. 15:06:28 4) Types improvements: 15:06:31 BP: #link https://blueprints.launchpad.net/manila/+spec/share-type-require-driver-mode 15:06:34 BP: #link https://blueprints.launchpad.net/manila/+spec/default-volume-type 15:06:43 status: work in progress 15:06:47 5) Manage/unmanage shares 15:06:50 BP: #link https://blueprints.launchpad.net/manila/+spec/manage-shares 15:06:53 gerrit: #link https://review.openstack.org/147495 15:06:57 status: finished, ready for review 15:07:03 6) level of visibility for shares 15:07:07 BP: #link https://blueprints.launchpad.net/manila/+spec/level-of-visibility-for-shares 15:07:10 gerrit: #link https://review.openstack.org/148853 15:07:13 status: ready for review 15:07:17 that's the main 15:07:36 so there's a lot there that's ready for review 15:07:59 I've been prioritizing K2 stuff over reviewing these so hopefully now I can get to all of those 15:09:01 I don't think any of the above is new, but now's your chance to ask if you want to know about something 15:09:35 i was curious about access level 15:09:52 is it going to stay the way it is for a while 15:09:52 that's just the read-only/read-write discussion we had a few weeks back 15:10:01 or were you thinking about adding more attributes 15:10:09 (or still digesting) 15:10:47 anything we add would need need support from backends 15:11:06 if some backends can't support a certain level then we can't really standardize it 15:11:14 i was thinking about the root-squash conversation 15:11:20 okdokey 15:11:22 read-only and read-write seem like the only universally understood access levels 15:11:51 root-squash needs a closer look -- it's a very common concept in the NFS world -- but I don't know if anyone can't support it 15:12:25 is anyone aware of a backend that can do NFS but can't do root squashing? 15:12:53 I do not know, need to investigate that 15:13:27 well this silence sounds promising 15:13:44 would not be able to do it on CIFS. Only on NFS. 15:13:54 yeah CIFS is totally different 15:14:19 CIFS actually has security in the protocol unlike NFS 15:14:33 * bswartz is joking.... 15:15:05 okay on to everyone's favorite topic 15:15:10 #topic manila midcycle meetup 15:15:26 So the dates are set 15:15:36 Feb 11 and Feb 12 15:16:12 so far nobody took me up on my offer to host people locally in RTP, which is what I expected 15:16:14 is there alreay an agenda? 15:16:47 I guess it would be useful to assign some times to the agenda 15:16:56 Since the meetup will be purely virtual, it should be better for those who can't join locally 15:17:06 oh yeah 15:17:09 #link https://etherpad.openstack.org/p/manila-kilo-midcycle-meetup 15:17:20 so the agenda is just a random list at the moment 15:17:39 we need to prioritize the items and do some scheduling 15:17:55 sounds good 15:18:11 will it be a web conference? or just irc? 15:18:16 honestly I've only ever done face-to-face meetups so I'm not experienced with the challenges we're going to face 15:18:35 I'm going to use a Google Hangout, with a telephone bridge 15:18:35 need mute-hammer =) 15:18:53 vponomaryov: lol +1 15:18:58 bswartz: ok let's try that... ;) 15:19:12 google hangouts is unfortunately limited in the number of people that can join, but a teleconference system can scale pretty well 15:19:25 and we've bridged the 2 before with relatively good success 15:19:28 voice qualality on google is really good though 15:19:54 so the main challenge for the meetup will be that, because nobody is traveling, we're all in different timezones 15:20:17 yep I won't be available all the time for instance 15:20:22 and the times I selected will be very difficult for some people (USA west coast and pacific ocean in particular) 15:20:34 so having timeslots for topics would be nice 15:20:41 which time zones are the developers mostly from? 15:20:49 USA and Europe? 15:20:55 dalgaaf: https://etherpad.openstack.org/p/manila-kilo-midcycle-meetup 15:20:57 dalgaaf: it all in here: https://etherpad.openstack.org/p/manila-kilo-midcycle-meetup 15:21:18 the plan is to duplicate some content in the "morning" and "afternoon" sessions so that people who can't join one can still participate 15:21:26 however I have no idea how well that will work 15:22:14 If there's stuff you'd like to talk about or hear about, please add it to the agenda 15:22:40 Personally I want to spend time on things that will matter during the rest of Kilo 15:23:04 bswartz: will you do the prioritization or should we do a voting? 15:23:34 please add your +1 to topics you want on the etherpad and I'll do my best 15:24:54 we can also discuss here if there's anything someone would really like to see/hear 15:25:36 the meetings that will be on wednesday can be duplicated, but what about thursday meeting? 15:25:37 My plan is to prepare a presentation on manila architecture (aimed at future developers and deployers) and to try it out on you guys 15:26:30 marcusvrn1: the "afternoon" session on Wednesday will have to cover anything from the "morning" sessions that people are interested in 15:26:59 I thought 12 hours would be the max we could all stand being on the phone 15:27:17 does it make even sense to duplicate ? 15:27:18 and I weighted the time towards the timezones that most people will be available 15:27:21 bswartz: ok! 15:27:40 maybe this format won't work out well 15:27:57 maybe we should add +1 and our names as well? 15:28:00 e.g. at the ceph developer Summit, which was completely virtual, we didn't duplicate 15:28:09 but I don't have a better idea given the international nature of the team 15:28:23 check e.g.: https://wiki.ceph.com/Planning/CDS/CDS_Giant_and_Hammer_%28Jun_2014%29 15:28:30 dalgaaf: I like that idea, but I'm wondering how spread out the ceph team is, timezone-wise 15:28:46 US, EMEA, Asia 15:29:10 the first day was US and EMEA mainly and the second day US and Asia 15:29:39 but you still can attend to both tracks if you are interested ;) 15:30:21 we won't have as much prepared presentation content for this summit 15:30:42 I want a lot of discussion to happen -- I think that's the main value of these meetups 15:30:52 and duplicating discussion is really hard to do 15:30:53 I'm simply not sure how it will work if you discuss a topic twice 15:31:29 yeah I don't think we'll duplicate discussions, but we can duplicate presentations and we can do a recap of earlier discussion for those that missed them 15:31:48 I expect we'll learn a lot by trying this out 15:32:09 and for L we can either plan it much further in advance and hope more people can travel, or try to shift to a format more like ceph's 15:32:28 is there a plan to record the sessions and put it e.g. on youtube? 15:32:48 I think we can get audio recorded fairly easily 15:33:22 dalgaaf: do you know who that works for the ceph summit.. is it just an option in google hangout? 15:33:46 I assume ... both google products 15:34:09 I can ask the ceph guys how it works 15:34:20 I hope to get my architecture presentation polished enough for a youtube video, but it won't be by next week 15:34:22 but writing imporant things down on etherpad works too IMHO 15:34:48 yeah we will absolutely take notes on etherpad 15:34:54 https://support.google.com/plus/answer/2553119?hl=en 15:35:24 I've been to a few Cinder meetups and I like very much how their format works, but they have the advantage of being mostly USA based, and people are able to travel 15:37:09 anything else about the meetup? 15:37:10 if the meetup is prepared 2 or 3 months in front it's may also possible to travel 15:37:15 and questions or things you'd like to see/hear? 15:37:29 yeah I hear you dalgaaf 15:37:43 for L we will schedule it before vancouver 15:38:34 and for those of you who don't know, I won't be attending Vancouver in person, due to the birth of my second child 15:38:58 congrats 15:38:58 so that will be an interesting challenge as well 15:39:02 oh congrats 15:39:13 congrats :) 15:39:14 at least vancouver isn't very far off from my home, timezone-wise 15:39:29 thank you :) 15:39:32 congrats 15:39:35 newborns don't care about time zones 15:39:42 heh 15:39:47 :) 15:40:05 well hopefully the baby will do me a favor and not be born until the Saturday after the conference 15:40:25 :) 15:40:26 but since these things are unpredictable, I plan to join remotely for as much as I can 15:41:07 #topic open discussion 15:41:20 so anything else for today? 15:41:39 tempest scenario testing 15:41:48 there is a survey to pick L name (and maybe name Ben's child?) 15:41:49 if not, we can all get started on reviews for K-3 and preparing any presentations for next week 15:41:59 markstur_: lol 15:42:02 I really hope we can start with them next week 15:42:12 mkoderer: yes! 15:42:29 mkoderer: I'm very interested in that topic 15:42:46 the first thing I'd like to see if a test plan that all interested parties can review 15:42:47 should we put them into contrib/ before we move them into tempest directly? 15:43:19 bswartz: ok. that should be possible :) 15:43:22 I don't really care where the code lives as long as it gets written and we can use it to do better end-to-end testing 15:43:38 bswartz: I remeber that I acutually created a wiki page about it.. let me search 15:43:44 tempest seems to be in flux at the moment anyways and it's not clear to me how it will shake out (or when) 15:43:53 yeah a link to the wiki would be good 15:44:06 #link https://wiki.openstack.org/wiki/Manila/specs/scenario-tests 15:44:12 vponomaryov: ty! 15:44:17 vponomaryov: thx :) 15:44:32 so this tables needs to get filled in with test plans 15:44:33 yeah it's empty but we need to fill it up 15:44:48 so we can make sure that we're covering everything we need to 15:45:08 once we can agree on the plan, we can create tasks and split them up amoung whoever's willing to work on them 15:45:15 bswartz: yep.. will do :) 15:45:26 thank you for leading on this mkoderer 15:45:53 for anyone not familiar with the issue -- the existing tempest tests don't yet exercise the data path for shares created by manila 15:46:30 so it's possible for serious bugs to creep into the network code and the driver share-access code and for tempest not to catch them 15:46:53 thus manual tests are still needed to make sure everything works 15:47:34 once we have these "scenario tests" merged and running in the gate, we can have much higher confidence that individual changes aren't breaking things 15:48:08 bswartz: it is not related to driver not covered by CI 15:48:16 any drivers 15:48:32 vponomaryov: that's another topic 15:49:00 obviously the gate will only run these tests using the generic driver -- which is significantly better than what we have now 15:49:15 Vendor CI will be required to properly test the drivers 15:49:24 and that's something I'd like to see for L 15:49:40 the experience of the Cinder team with CI has been very interesting 15:49:56 it's been painful for many, but ultimately it's been very positive 15:50:57 forcing vendors to test their drivers in an automated way really improves quality and increases community participation 15:51:53 some might argue that vendors should be testing their code anyways, but that's been proven to be untrue 15:52:24 bswartz: 95% of developers do not test their code =) 15:52:35 heh 15:52:36 vponomaryov: +1 15:53:10 alright I think that's it for today 15:53:28 oh I'll put scenarios tests on the meetup agenda 15:53:31 time to start socializing the cap ex 15:53:33 I want to talk about that some more 15:53:35 to run CI 15:53:52 okay I can put CI on the agenda to, so it's not a big surprise to anyone when we propose it for L 15:54:08 thanks everyone and see you next week 15:54:09 there will be some lead time i would expect 15:54:12 so sounds good 15:54:54 #endmeeting