16:01:55 <jgriffith> #startmeeting cinder
16:01:55 <openstack> Meeting started Wed Sep 18 16:01:55 2013 UTC and is due to finish in 60 minutes.  The chair is jgriffith. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:01:56 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
16:01:58 <openstack> The meeting name has been set to 'cinder'
16:02:06 <bswartz> hi
16:02:12 <med_> \o
16:02:16 <zhiyan> \o/
16:02:16 <xyang_> hi
16:02:21 <thingee> o/
16:02:24 <eharney> hi
16:02:25 <jgriffith> Hey there everyone
16:02:29 <jgriffith> Looks like a good group
16:02:37 <jgriffith> missing avishay and winston
16:02:45 <jgriffith> but we'll move on and have a short meeting
16:02:59 <jgriffith> I have a couple of things I wanted to remind everyone
16:02:59 <jungleboyj> A short meeting?  Bwah ha ha!
16:03:02 <jgriffith> caitlin-nexenta: ping
16:03:10 <jgriffith> jungleboyj: hehe... well, we can always have goals
16:03:11 <guitarzan> without those two it's possible! :D
16:03:15 <caitlin-nexenta> jgriffth: yes?
16:03:21 <jungleboyj> jgriffith: +2
16:03:32 <jgriffith> caitlin-nexenta: wanted to make sure you were here for this first topic since Victor doesn't do IRC
16:03:43 <jgriffith> #topic Stop submitting your feature patches
16:04:01 <jgriffith> As you've noticed if you submit a new feature patch I'm just going to -2 it anyway until we open up icehouse
16:04:16 <jgriffith> Please refrain from continuing to propose new feature patches for the time being
16:04:19 <zhiyan> jgriffith: when it will be open?
16:04:30 <jgriffith> zhiyan: depends on when we get RC1 cut
16:04:31 <hemna> jgriffith, shouldn't that go for anything that doesn't have a bug associated with it ?
16:04:37 <jgriffith> hemna: yes
16:04:59 <jgriffith> hemna: I should've specified but assumed process of elimination :)
16:05:02 <hemna> I've seen a few reviews that are nothing more than code cleanups....no bugs associated
16:05:04 <zhiyan> jgriffith: got it
16:05:09 <bswartz> jgriffith: so would now be a bad time to submit another file-shares patch?
16:05:23 <bswartz> jk
16:05:26 <jgriffith> hemna: fell free to -2 them unless they're minimal and worth putting in
16:05:39 <jgriffith> but at this point we want to try and limit any risk of regression/failures
16:05:56 <jgriffith> bswartz: HA!!!
16:06:05 <jgriffith> bswartz: you are funny this morning
16:06:15 <jgriffith> bswartz: you're covering for kmartin ?
16:06:29 <bswartz> yes
16:06:30 <hemna> ok
16:06:45 <kmartin> I'm here
16:06:46 <hemna> that's what I was hoping you'd say.  I'll -2 those now then.  thanks
16:06:46 <jgriffith> so anyway, just wanted to say.  Please refrain for now as it just clutters the gates and the review dashboard
16:06:52 <jgriffith> kmartin: you have help today :)
16:07:05 <jgriffith> #topic testing
16:07:14 <jgriffith> We need all hands on deck for testing
16:07:23 <jgriffith> the more we find and fix now the better
16:07:36 <jgriffith> I'd like to get the RC cut and move on to icehouse as soon as possible
16:07:43 <hemna> +1
16:07:48 <jgriffith> the only thing standing in the way for icehouse work is getting the RC cut
16:07:54 <jgriffith> so you see the priority :)
16:07:58 <hemna> when is the RC ?
16:08:04 <jgriffith> hemna: depens on when we're ready
16:08:08 <hemna> ok
16:08:08 <kmartin> nice, can I get help making PowerPoint slides?
16:08:17 <jgriffith> hemna: the way it works is we close out the issues for RC1
16:08:19 <bswartz> wait we still have 7 weeks to the developer summit
16:08:19 <jgriffith> we test
16:08:26 <jgriffith> see how the bug trajectory goes
16:08:32 <jgriffith> repeat until there are no bugs
16:08:38 <jgriffith> rc1, 2, 3
16:08:45 <jgriffith> if we go past 3 we're doing something wrong
16:08:53 <hemna> that's bad mmmkay
16:08:57 <bswartz> what kind of icehouse work should be happening between now and then? what about havana DOCS etc?
16:09:11 <jungleboyj> kmartin: I can spell check for you. ;-)
16:09:12 <jgriffith> bswartz: yes, docs as well
16:09:12 <hemna> ew, that 4 letter word
16:09:24 <jgriffith> bswartz: none right now
16:09:33 <jgriffith> bswartz: and none depending on where we are in a couple weeks
16:09:52 <jgriffith> bswartz: the window is there for an early start but that doesn't mean we'll be able to or should use it
16:10:13 <jgriffith> bswartz: by the time we get H ready and get sessions set up I don't think 7 weeks is as long as some may think
16:10:44 <thingee> worked with a deployer on getting cinder running. made me take notes on things I found missing in the docs.
16:10:58 <jgriffith> thingee: nice!
16:11:06 <jgriffith> thingee: publish it somewhere and we should all pitch in
16:11:40 <thingee> it's not a big list, but it could probably be expanded once people see it and think about it.
16:12:19 <jgriffith> thingee: yeah, hoping mind share will help
16:12:30 <jgriffith> thingee: I'd really like to excel on the docs side if we could
16:12:47 <jgriffith> G was a HUGE improvement
16:12:56 <jgriffith> we can/should build on that momentum
16:12:58 <jungleboyj> jgriffith: I am happy to help with that as a relative newbie.
16:13:02 * med_ is hoping to do some doc-ing that gets approved
16:13:10 <jgriffith> med_: :)
16:13:17 <jungleboyj> jgriffith: Can bring fresh eyes and questions to it.
16:13:31 <jgriffith> coolness
16:13:36 <jgriffith> speaking of summit....
16:13:42 <jgriffith> #topic design summit
16:13:44 <jgriffith> http://summit.openstack.org/
16:13:52 <jgriffith> IMO this is a great list so far
16:14:37 <jgriffith> I suspect all of these will be approved
16:14:44 <jgriffith> thanks to folks for jumping on these early
16:14:58 <jgriffith> with that...
16:15:00 <jgriffith> I think I'm done
16:15:02 <kmartin> still room for a few more
16:15:14 <jgriffith> kmartin: yes, still room which is great
16:15:15 <hemna> jgriffith, we may have another BP and session proposal on the way
16:15:21 <jgriffith> hemna: bring it on
16:15:41 <jgriffith> Ok, anybody have anything they want to add/ask regarding sessions?
16:15:42 <kmartin> when is the cutoff date for submitting them?
16:15:44 <eharney> i probably do too
16:16:02 <jgriffith> kmartin: you've got some time, but I guess considering the legal process you should move quickly :)
16:16:09 <med_> heh
16:16:09 <jgriffith> kmartin: I don't have a cut-off date right now
16:16:30 <jgriffith> kmartin: typically we go up until the 11'th hour on session organization etc
16:16:40 <kmartin> I worked around those issues, I can do whatever I want for the next 3 releases
16:16:43 <jgriffith> kmartin: although it would be nice to NOT do that this time
16:16:51 <jgriffith> kmartin: Nice!!!
16:16:55 <jgriffith> kmartin: buy me a pony!
16:17:09 <hemna> or a free hpcs account? :P
16:17:22 <jgriffith> hemna: Rax gave me one of those :)
16:17:27 <jgriffith> hemna: or at least a credit
16:17:30 <hemna> nice!
16:17:45 <jgriffith> Ok...
16:17:45 <kmartin> jgriffith: more like a lottery ticket for the hoops I had to go though
16:17:52 <jgriffith> kmartin: hehe
16:17:59 <jgriffith> kmartin: I can only imagine
16:18:04 <jgriffith> #topic open-discussion
16:18:12 <jgriffith> anything on anybody's mind?
16:18:23 <hemna> so the manager fix?
16:18:31 <jgriffith> hemna: :)
16:18:36 <hemna> heh
16:18:36 <med_> so the devref docs
16:18:43 <hemna> if no one else wants the floor
16:18:45 <med_> is that doable for H or am I thinking I now?
16:18:55 <jgriffith> med_: def H
16:18:58 <med_> nodz.
16:19:02 <jgriffith> med_: docs are special category :)
16:19:09 <jgriffith> hemna: take it
16:19:12 * med_ feels special
16:19:21 <jgriffith> med_: *is* very special
16:19:24 <hemna> ok, so I have https://review.openstack.org/#/c/46843/ outstanding
16:19:35 <hemna> It tries to enforce the driver initialization flag
16:19:42 <hemna> which the manager was basically all but ignoring
16:19:54 <hemna> I have some jenkins issues to work out today
16:20:17 <hemna> but the idea is to have the manager enforce the driver being initialized properly.
16:20:48 <hemna> I need to look more at the backup manager as well and make sure it's being enforced there as well
16:21:25 <kmartin> jgriffith:  Regarding the cinderclient changes for the qos support. Is that going to get in H? https://review.openstack.org/#/c/46979
16:22:05 <hemna> jgriffith, you mentioned you had some feedback for me on this yesterday, you can ping me outside the meeting if you like
16:22:24 <hemna> I presume we want this to make RC
16:23:31 <hemna> ?
16:24:55 <kmartin> seems that jgriffith moved the meeting to the #openstack-cinder?
16:25:07 <hemna> crickets
16:25:11 <bswartz> kmartin: yes
16:25:36 <hemna> chirp
16:25:37 <hemna> chirp
16:25:37 <jgriffith> DERP!!!
16:25:48 <jgriffith> I just started my rant in #cinder
16:25:50 <kmartin> and jgriffith is back
16:25:52 <hemna> lol
16:26:01 <jgriffith> sorry bout that
16:26:01 <hemna> jgriffith, ok, so I'm on board w/ you on the manager not breaking
16:26:05 <jgriffith> anyway
16:26:06 <hemna> I think that's great
16:26:13 <hemna> and I'm enforcing that in my patch
16:26:16 <jgriffith> Ok, we'll get your patch looked at and move on
16:26:32 <hemna> all I did is just add some logging of the caught exception.
16:26:33 <jgriffith> hemna: I agree that there needs to be a second half to the change
16:26:41 <jgriffith> hemna: perfecto
16:26:44 <hemna> and then enforcing the initialized flag in the rest of the manager.
16:26:55 <caitlin-nexenta> henma: could you save me a lot of reading, where are the QOS specs enforced?
16:27:04 <kmartin> jgriffith:  Regarding the cinderclient changes for the qos support. Is that going to get in H? https://review.openstack.org/#/c/46979
16:27:25 <jgriffith> kmartin: it has to :)
16:27:34 <hemna> ok coolio.   I'll fix the jenkins issues today and put up another patch on it.
16:27:42 <jgriffith> kmartin: I'm waiting on that from winston and avishay's migration changes
16:27:44 <DuncanT-> caitlin-nexenta: Either at the hypervisor or the backend, depending on which you set when you create the limit
16:27:49 <jgriffith> then I plan to push to PyPi
16:28:09 <kmartin> jgriffith: kind of needs land before I start making the changes in the driver
16:28:12 <caitlin-nexenta> DuncanT: thanks.
16:28:17 <jgriffith> kmartin: huh?
16:28:24 <jgriffith> kmartin: you messin with me again?
16:28:37 <caitlin-nexenta> That's something I have to check. It's not the most visible of constraints for backend implementers.
16:29:23 <zhiyan> jgriffith: and what about cinderclient change for r/o-volume support, do you think we can landing it concurrently? https://review.openstack.org/#/c/44672/ and https://review.openstack.org/#/c/45171/
16:29:23 <jgriffith> kmartin: ?? chirp chirp chirp
16:29:31 <jgriffith> zhiyan: oh about tht
16:29:32 <jgriffith> that
16:29:53 <kmartin> jgriffith: sorry I was in the cinder room
16:29:54 <jgriffith> I thought last week we all decided that since it wasn't supported in Nova we were going to disable it in the policy?
16:30:00 <jgriffith> kmartin: HA!
16:30:23 <jgriffith> zhiyan: in which case I don't want to put something in the client that can't actually be used yet
16:30:28 <jgriffith> zhiyan: or am I missing something?
16:30:29 <zhiyan> jgriffith: yes, i know. but those two are for cinderclient.
16:30:43 <jgriffith> zhiyan: sure, but why expose them if you can't use them?
16:31:51 <zhiyan> jgriffith: actually those be used by nova...humm, yes, actually i just want to try land them concurrently to save time, and nova server side change can be landing asap when FF finished..
16:32:30 <jgriffith> zhiyan: I'd prefer to see it as a WIP and as soon as we release RC we can update
16:32:51 <kmartin> jgriffith: it would be difficult to test qos feature if I can't set the qos_specs and relate them to volume types
16:32:54 <zhiyan> jgriffith: ok.
16:32:54 <jgriffith> zhiyan: makes it cleaner for pkg maintainers/builders for H IMO
16:33:25 <jgriffith> kmartin: yeah, I've been using curl and tried to get postman workign again :)
16:33:33 <zhiyan> jgriffith: make sense. thanks. will mark it wip until FF finish.
16:33:45 <jgriffith> kmartin: the problem I see is that since we're well beyond FFE we're all kinda stuck WRT to the new stuff around QoS
16:33:52 <jgriffith> zhiyan: thank you
16:34:22 <jgriffith> kmartin: the onslaught of driver/bug's changes would be highly annoying and subject to all sorts of problems IMO
16:34:28 <jgriffith> kmartin: but that being said
16:34:43 <kmartin> jgriffith: ok, it is pretty late so I can wait and do it in Icehouse
16:34:44 <jgriffith> kmartin: let's get an update from winston or get it in our selves and go from there
16:34:53 <kmartin> jgriffith: ok
16:34:53 <dosaboy> jgriffith: did you have any thoughts about brick?
16:35:02 <hemna> dosaboy, what about brick ?
16:35:15 <dosaboy> you mentioned you had spome concerns and wanetd to review where it was going?
16:35:16 <jgriffith> kmartin: I'd like to get a gauge on everybody's needed activity on that once we have the client changes in
16:35:18 <caitlin-nexenta> How about somebody defining it.
16:35:34 <hemna> ah yes, that was my todo last week wasn't it.
16:35:35 <hemna> d'oh
16:35:40 <dosaboy> hemna: so i'm planning to add some rbd bits to brick
16:35:52 <jgriffith> caitlin-nexenta: shared library of commone block storage related functions
16:35:54 <kmartin> sounds good, I think only a few drivers actually support QOS
16:36:07 <dosaboy> but I understand that we want to avoid brick becoming a holdall for everybodies common driver code
16:36:12 <hemna> jgriffith, initiator and target right ?
16:36:24 <jgriffith> dosaboy: yeah, which is what people are rapidly trying to do
16:36:28 <hemna> dosaboy, correct, it's not a driver repository
16:36:30 <jgriffith> hemna: yes
16:36:37 <jgriffith> hemna: and base LVM functionality
16:36:38 <med_> is brick part of cinder or a separate library (and is it used in other projects (nova, etc))
16:36:46 <caitlin-nexenta> jgriffith: that's a non-definition. *which* common functions?  With that definition you should call it "kitchen sink"
16:36:50 <jgriffith> part of cinder right now that will be a lib for I
16:36:58 <med_> nod
16:36:59 <dosaboy> med_: afaik brick is intended to become an exportable lib
16:37:00 <jgriffith> caitlin-nexenta: it is a definition, you just don't like it
16:37:10 <jgriffith> everybody hold up for a second..
16:37:18 <jgriffith> So the idea was/is simply this:
16:37:26 * dosaboy digs out a bp
16:37:33 <jgriffith> Currently there is a bunch of duplicate code between nova and cinder
16:37:55 <jgriffith> This duplicate code I'm referring to is around handling initiators and targets in iscsi
16:38:03 <jgriffith> as well as connection code for FibreChannel etc
16:38:18 <jgriffith> In addition, both Nova and Cinder do a good deal with LVM
16:38:38 <jgriffith> One aspect of Brick was intended to consolidate those things
16:38:49 <jgriffith> NOT for everybody to put their driver in brick
16:39:00 <dosaboy> jgriffith, hemna: so right now I have https://blueprints.launchpad.net/cinder/+spec/add-rbd-support-to-cinder-brick
16:39:07 <dosaboy> it is looseluy defined
16:39:16 <dosaboy> i see there is a ne task in there too
16:39:35 <dosaboy> but basically, since rbd support is now in nova + glance + cinder
16:39:49 <jgriffith> dosaboy: yes, that would make it a candidate here
16:39:54 <dosaboy> i think there is defo scope for some common rbd code going int
16:39:59 <dosaboy> good good
16:40:23 <hemna> my eventual plan is to take the kvm libvirt volume driver code and put them all as connectors in brick
16:40:23 <jgriffith> dosaboy: but this has become a bit muttled and we need clearer distinctions and long term goals
16:40:30 <hemna> I have iscsi and FC now
16:40:34 * dosaboy has yet to actually find a chance to code it
16:40:44 <dosaboy> i hope to have something proto'd by I
16:41:03 <dosaboy> hemna: nod
16:41:25 <jgriffith> We'll regroup on this at the summit by the way
16:41:34 <caitlin-nexenta> Some drivers getting into brick will create an incentive for all drivers to get their code their as well. A utility library should be driver neutral. I'm seeing far too much about LVM and RBD.
16:41:52 <jgriffith> caitlin-nexenta: do you have a copy of your driver in Cinder and in Nova?
16:42:22 <jgriffith> caitlin-nexenta: I agree with you
16:42:33 <dosaboy> caitlin-nexenta: rbd is actually a protocol not just a driver
16:42:34 <caitlin-nexenta> The original would still be in nova, but we've been updating in Cinder.
16:42:46 <jgriffith> caitlin-nexenta: original?
16:42:49 <dosaboy> which is natively supported alongside iscsi in some hypervisors
16:42:53 <jgriffith> caitlin-nexenta: what are you talking about there?
16:43:13 <jgriffith> dosaboy: but there's a better/different way to handle that IMO
16:43:21 <caitlin-nexenta> Actually I'm not sure if it is still there. There was a nexenta driver in nova-volume before Cinder.
16:43:25 <dosaboy> jgriffith: pray tell
16:43:28 <jgriffith> caitlin-nexenta: it's not
16:43:37 <jgriffith> caitlin-nexenta: we did what was called a volume-ectomy
16:43:56 <med_> heh
16:43:59 <caitlin-nexenta> ok, it would still work - but it's nice not to have to worry about it.
16:44:02 <jgriffith> all of the block device drivers were removed from Nova
16:44:07 <jgriffith> caitlin-nexenta: it woudl not
16:44:16 <hemna> *sigh*
16:44:26 <jgriffith> caitlin-nexenta: the Nova code has been completely redone
16:44:27 <jgriffith> anyway
16:44:30 <jgriffith> Ok...
16:44:36 <jgriffith> Here's my take for now:
16:44:41 <jgriffith> NO drivers go in brick
16:44:43 <jgriffith> period
16:44:51 <hemna> +2
16:44:52 <kmartin> jgriffith: +3
16:45:02 <jgriffith> connectors, helpers etc we'll address as needed
16:45:12 <dosaboy> agreed
16:45:21 <caitlin-nexenta> And those should be defined in ways that are driver neutral.
16:45:50 <hemna> caitlin-nexenta, they already are, so I'm not sure what your complaints are about at this point.
16:46:07 <hemna> next topic ?
16:46:20 <DuncanT-> caitlin-nexenta: brick is not designed to be driver neutral... it is a place for useful common code, not a place that many drivers need to touch
16:47:19 <jgriffith> ok I think we're done for now
16:47:22 <caitlin-nexenta> THen why isthe title "Add RBD support to cinder brick"
16:47:23 <jgriffith> no need to rat-hole
16:47:25 <hemna> anngentle wanted us to see if anyone was going to write docs for volume encryption
16:47:38 <jgriffith> caitlin-nexenta: who cares, it's not approved, it's not implemented
16:47:45 <dosaboy> it is a bad title
16:47:47 <jgriffith> caitlin-nexenta: let's not worry about things that don't currently exist
16:47:56 <caitlin-nexenta> ok
16:48:06 <jgriffith> DuncanT-: +1 by the way
16:48:07 <jgriffith> Ok
16:48:13 <jgriffith> let's adjurn
16:48:24 <jgriffith> so much for the 15 minute meeting I had envisioned :(
16:48:30 <jgriffith> #endmeeting cinder