15:00:23 <bswartz> #startmeeting manila
15:00:24 <openstack> Meeting started Thu Oct 15 15:00:23 2015 UTC and is due to finish in 60 minutes.  The chair is bswartz. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:26 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:00:28 <openstack> The meeting name has been set to 'manila'
15:00:31 <cknight> Hi
15:00:32 <ganso> Hello
15:00:33 <bswartz> hello all
15:00:38 <csaba> hi
15:00:38 <Zhongjun> hello
15:00:41 <vponomaryov> hi
15:00:51 <bswartz> so today there is no real agenda
15:00:56 <cfouts> hi
15:00:56 <bswartz> it will be a short meeting
15:00:58 <xyang1> Hi
15:01:03 <bswartz> I have 2 things to mention
15:01:17 <bswartz> #topic liberty release
15:01:20 <dustins> \o
15:01:53 <bswartz> liberty release is today -- there was an RC3 on tuesday to fix a critical bug which carried a bunch a gate-block-fixes back to liberty
15:02:05 <tbarron> hi
15:02:30 <bswartz> the main thing for you guys is that as you find bugs in mitaka, if they need backporting the correct tag is now "liberty-backport-potential"
15:03:09 <vponomaryov> bswartz: should we set also set "kilo-*" and "juno-*" tags?
15:03:24 <bswartz> thanks to all to tested the Liberty release and found and fixed bugs, I think it was very successful
15:04:07 <bswartz> also thanks to those who implemented new features and very special thanks to those who did the hard work of setting up CI systems for vendor drivers -- I know that work is not easy but I do believe it pays off
15:04:34 <bswartz> vponomaryov: stable/juno should go away very soon, and stable/kilo will only be open for security fixes
15:04:51 <bswartz> that's according the stable maintenance policy
15:04:53 <vponomaryov> bswartz: ah, ok, thanks
15:05:26 <bswartz> if there's any critical bugfixes that need backported to kilo, let's do it before the weekend
15:05:52 <bswartz> the other topic is
15:05:57 <bswartz> #topic mitaka design summit
15:06:05 <bswartz> #link https://wiki.openstack.org/wiki/Design_Summit/Mitaka/Etherpads#Manila
15:06:32 <bswartz> those are the session I selected -- it's still possible to make changes though if someone has strong feelings
15:06:53 <bswartz> xyang1: I didn't get a chance to talk to you earlier about your session proposal: "minimum required features"
15:07:43 <xyang1> Can we talk about that at a work session?  Or we are out of slots
15:07:46 <bswartz> xyang1: I didn't really understand what the point of that session would be and it didn't sound to me like a full 40 minute topic
15:08:01 <bswartz> we can if you think it needs a full 40 minutes and if you want to present something
15:08:08 <bswartz> I'll move something else to make room
15:08:11 <xyang1> Not need a full session, but we should decide on it
15:08:18 <bswartz> the alternative is to cover it a the contributor meetup
15:08:35 <bswartz> I know you have conflict between the manila contributor meetup and the cinder all-day meetup
15:08:38 <xyang1> I may not be able to make it on Friday
15:08:43 <xyang1> Right
15:09:15 <bswartz> xyang1: is it something we could cover during a weekly meeting? such as today or next week?
15:09:24 <bswartz> or does it require us all to be in a room together?
15:09:35 <xyang1> That is fine too
15:09:51 <bswartz> okay I want to make sure it gets the time it deserves
15:09:52 <xyang1> We need to make a decision and put on wiki
15:09:57 <bswartz> it just didn't feel like a full 40 minutes to me
15:10:16 <xyang1> I don't think it needs 40 minutes
15:10:18 <bswartz> ganso: you had an etherpad with minimum required features
15:10:28 <bswartz> (I think it was ganso)
15:10:30 <ganso> bswartz: yes
15:10:38 <xyang1> However we spent lots of time talking about snapshot
15:10:42 <bswartz> did that ever get converted into a dev doc in the manila tree?
15:10:53 <bswartz> ganso: ^
15:10:54 <ganso> bswartz: no
15:11:03 <bswartz> okay well that's something we should probably do
15:11:07 <xyang1> Maybe this can be done via code reviews
15:11:16 <cknight> xyang1: +1
15:11:17 <vponomaryov> +1
15:11:21 <ganso> #link https://etherpad.openstack.org/p/manila-minimum-driver-requirements
15:11:22 <bswartz> xyang1: I made one the fishbowl sessions to discuss new snapshot semantics
15:11:40 <bswartz> because that topic will probably be of great interest to many people
15:11:55 <bswartz> I'm not sure if there are any other features which will invite a lot of discussion
15:11:59 <ganso> bswartz: I suggested merging both subjects, but that would be up to xyang1
15:12:43 <bswartz> I'm in favor of managing that through code review
15:12:53 <ganso> as far as I remember, during midcycle meetup, one of the topics was minimum required features and snapshot was 80% of the discussion
15:13:04 <bswartz> if we discover that there is another feature which invites a lot of discussion, then we can set aside time for those discussions
15:13:23 <bswartz> but I would bet that there's nothing else contentions
15:13:27 <cknight> ganso: Yes, but that was whether basic snapshots would be part of the minimum set.  We did agree they would not be.
15:13:36 <bswartz> contentious*
15:13:57 <xyang1> ganso: can you submit a patch before the summit
15:14:08 <bswartz> +1 for dev doc patch before summit
15:14:08 <ganso> xyang1: yes
15:14:13 <bswartz> ganso: thanks
15:14:29 <bswartz> okay that's all I had for today
15:14:32 <bswartz> #topic open discussion
15:14:36 <xyang1> So we can start reviewing and make sure everyone agrees
15:14:54 <xyang1> If there are conflicts, we discuss at the summit
15:15:14 <bswartz> anyone else have something for this week?
15:15:17 <Zhongjun> I made a small change in Manila QoS design.
15:15:33 <Zhongjun> Manila QoS add a few common capabilities in qos-create command.
15:15:49 <bswartz> thanks Zhongjun
15:15:57 <Zhongjun> such as: manila qos-create <name> [--max_kbps <max_kbps>] [--min_kbps <min_kbps>] [--max_iops <max_iops>] [--min_iops <min_iops>] <key=value> [<key=value> ...]
15:16:12 <Zhongjun> Is it reasonable for that?
15:16:27 <bswartz> I suggest people review the QoS design before tokyo -- it's on the agenda to be discussed during the contributor meetup
15:17:03 * bswartz reminds himself to update the etherpads with topics
15:17:08 <Zhongjun> #link: https://wiki.openstack.org/wiki/Manila/QoS
15:17:54 <bswartz> Zhongjun: that's reasonable, it looks similar to what cinder has
15:18:16 <bswartz> I'm not convinced we want to copy cinder's design exactly, but if that's what others want then I would be okay with it
15:18:47 <bswartz> In particular I'm unclear on what "min" means
15:19:09 <bswartz> I want to hear from the vendors that can implement a min_iops and an min_kpbs what it means for their systems
15:19:20 <Zhongjun> min -> minimum
15:19:44 <bswartz> yes but how can it be enforced -- is it a hard limit or a soft limit?
15:19:55 <jcsp> for ceph, it would be useful to have a limit on the metadata op rate, separate from the data IOPS/bandwidth
15:20:07 <jcsp> we don't have QoS in there yet, but when we add it there'll certainly be a separate between data and metadata qos.
15:20:10 <bswartz> it seems it would have to be soft because if the whole system bogs down then nobody will get the minimums they were promised
15:20:41 <cknight> jcsp: The design allows for backend-specific QoS keys.  But common ones should be, well, common!
15:20:47 <bswartz> presumably vendors who implement qos minimums have a way to ensure their systems don't bog down, but I don't understand how that works
15:21:03 <Zhongjun> yes, for huawei, it would have to be soft
15:21:38 <bswartz> Zhongjun: it could just be a matter of documentation, to make sure users are clear on what it means
15:22:18 <bswartz> okay anything else for today?
15:22:19 <Zhongjun> bswartz: ok, I will add the detail about that.
15:23:11 <bswartz> alright thanks everyone
15:23:18 <bswartz> one more week to tokyo!
15:23:32 <bswartz> #endmeeting