13:00:03 <alex_xu> #startmeeting nova api
13:00:04 <openstack> Meeting started Wed Oct 18 13:00:03 2017 UTC and is due to finish in 60 minutes.  The chair is alex_xu. Information about MeetBot at http://wiki.debian.org/MeetBot.
13:00:05 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
13:00:08 <openstack> The meeting name has been set to 'nova_api'
13:00:13 <alex_xu> who is here today?
13:00:14 <takashin> o/
13:00:19 <gmann> o/
13:00:51 <alex_xu> #topic specs
13:01:10 <alex_xu> #link https://review.openstack.org/508101
13:01:18 <alex_xu> gmann: how was that?
13:01:59 <gmann> yea, i am still not sure about instance_name attr - https://review.openstack.org/#/c/508101/5/specs/queens/approved/api-extensions-policy-removal.rst@91
13:02:01 <gmann> #link https://review.openstack.org/#/c/508101/5/specs/queens/approved/api-extensions-policy-removal.rst@91
13:02:02 <alex_xu> looks like a -1 from jichenjc
13:02:24 <gmann> because of OS-EXT-SRV-ATTR:instance_name
13:02:39 <gmann> should we consider this admin related and keep the policy for this
13:03:02 <gmann> may be you know the background as you remembered it was for ec2
13:05:44 <gmann> sdague:  johnthetubaguy any feedback on 'OS-EXT-SRV-ATTR:instance_name' ?
13:05:56 <gmann> #link https://review.openstack.org/#/c/508101/5/specs/queens/approved/api-extensions-policy-removal.rst@91
13:07:00 <johnthetubaguy> what is it again?
13:07:47 <gmann> johnthetubaguy: in this spec, i am proposing the deprecation of few policy which are related to extensions as discussed in PTG
13:07:49 <gmann> #link v
13:07:53 <gmann> #link https://review.openstack.org/508101
13:07:55 <johnthetubaguy> actually, I see, the API docs say it should be for admins only, that makes sense I guess
13:08:16 <johnthetubaguy> "OS-EXT-SRV-ATTR:instance_name": "instance-00000001",
13:08:21 <johnthetubaguy> is the example in the API docs
13:08:28 <gmann> so that policy control many other attribute also like user_data which does not seems admin one
13:08:29 <johnthetubaguy> that is of no use to an end user really
13:08:37 <alex_xu> ah, it is the name store in the libvirt xml
13:08:54 <johnthetubaguy> yeah, its the name shown in the underlying hypervisor
13:09:10 <gmann> ok
13:09:17 <johnthetubaguy> user_data is just really big, I suspect thats what it wasn't returned
13:09:25 <gmann> johnthetubaguy: i see
13:09:36 <gmann> i remember of seeing that name with virsh list
13:10:10 <alex_xu> I remember someone filed bug about show the user_data for the normal user
13:12:05 <gmann> ok,
13:12:31 <alex_xu> emm...I can't find that bug
13:12:33 <johnthetubaguy> I really worry about exposing new things all of a sudden, I wasn't in that PTG discussion I guess, so not sure on the context
13:12:43 <gmann> johnthetubaguy: alex_xu should we keep all attribute listed @91 as it is. i mean no change in policy ?
13:12:53 <johnthetubaguy> so there is a whole heap of deprecation's we should totally do
13:13:25 <gmann> johnthetubaguy: main idea is to deprecate the policy which were more related to and added when api extensions were added
13:13:27 <takashin> alex_xu: this one? https://bugs.launchpad.net/nova/+bug/1670978
13:13:28 <openstack> Launchpad bug 1670978 in OpenStack Compute (nova) "most of extended server attributes returned in 2.3 api versions should not require admin role" [Wishlist,In progress] - Assigned to jichenjc (jichenjc)
13:13:53 <johnthetubaguy> I think we need to work out how to make the policy change on those attributes, it almost feels like we should have a microversion for the change
13:13:58 <gmann> johnthetubaguy: for example polucy 'os_compute_api:os-extended-availability-zone' @70
13:14:17 <alex_xu> takashin: thanks
13:14:37 <gmann> johnthetubaguy: microversion for policy change?
13:15:03 <johnthetubaguy> depends how you look at this, to 90% of the people using our API we are adding new attributes
13:15:22 <sdague> how did instance_name ever get into things?
13:15:54 <sdague> I wonder why ec2 needs those
13:16:17 <johnthetubaguy> sdague: thats probably more to do with admins wanting to know what instance name to look at in their system
13:17:05 <johnthetubaguy> oh... ec2 uses instance-00000001 as its name, rather than the user one, back in the day
13:17:45 <johnthetubaguy> (...thinking back to when I was using euca tools because python-novaclient wasn't finished)
13:17:59 <sdague> oh
13:18:17 <sdague> gotcha, though that's weird given that it's not unique :)
13:18:19 <alex_xu> oh, sorry, the 2.3 microversion is for ec2, but the name isn't in the list
13:18:48 <johnthetubaguy> oh, and you can break it by chaning your config
13:19:07 <gmann> yea by CONF.instance_name_template
13:20:42 <johnthetubaguy> Just to baseline, I guess we all agree on dropping the crazy policies that default to allow, but just randomly kill parts of an ex-extension?
13:21:22 <johnthetubaguy> (by drop I mean deprecate...)
13:21:37 <gmann> yea, default to allow all one are main candidates
13:21:46 <gmann> yea by deprecation
13:22:28 <gmann> so in that list in spec, only ''os_compute_api:os-extended-server-attributes'' is admin one only all other are default to all
13:22:50 <gmann> i will remove this policy ('os_compute_api:os-extended-server-attributes') from proposal
13:23:46 <gmann> ^^ is ok?
13:24:48 <johnthetubaguy> I think we keep that problem separate, while we think about the best way forward
13:24:56 <johnthetubaguy> (i.e. don't block the other good progress)
13:25:04 <johnthetubaguy> so +1 from me
13:25:11 <gmann> yea
13:25:29 <gmann> ll update spec tomorrow.
13:25:37 <alex_xu> not sure johnthetubaguy and sdague is knowing the background, we want to keep the policy for the admin-only attributes since the admin may want to configure those attrs for other role like admin
13:25:59 <alex_xu> does sound make sense?
13:26:09 <johnthetubaguy> yeah, we totally have to keep the admin policy either way
13:26:26 <johnthetubaguy> the question is about what attributes it applies (and if that changes depending on the microversion)
13:28:17 <gmann> sure. let's target non 'admin only' first
13:29:12 <gmann> johnthetubaguy: sdague alex_xu thanks for feedback
13:29:24 <alex_xu> johnthetubaguy: sorry, the question is reference to the problem separation?
13:30:28 <johnthetubaguy> probably ignore that comment, just thinking about the admin-only instance_name and user_data
13:30:51 <alex_xu> ah, i see
13:31:05 <alex_xu> johnthetubaguy: thanks
13:32:00 <alex_xu> gmann: fyi, tomorrow is the spec-freeze
13:32:31 <gmann> alex_xu: :) yea, i can update this now quickly
13:32:32 <alex_xu> last chance is probably the tomorrow night
13:32:55 <alex_xu> ok...anything other spec people want to bring up?
13:33:16 <alex_xu> gmann: thanks
13:33:28 <gmann> nothing from my side.
13:34:00 <takashin> nothing
13:34:23 <alex_xu> ok, so I guess nothing for the open also
13:34:58 <alex_xu> let us close the meeting, thanks all!
13:35:00 <gmann> yea from me
13:35:03 <gmann> alex_xu: thanks
13:35:07 <alex_xu> #endmeeting