Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2017-01-05
12:21:57 sfinucan True. What about the 'driver' one?
12:22:47 johnthetubaguy sfinucan: well I don't quite get why we deprecated that I suppose, I guess we will kill the no op driver, but I suspect folks use that
12:22:53 gibi diga_: yes. how can I help you?
12:23:18 diga_ gibi: jaypipes told me to work with you on nova-placement-notifications
12:23:35 johnthetubaguy sfinucan: I thought it was just going to move to stevedore, in which case, the setting remains, its just we do a fallback on the values used
12:24:06 diga_ gibi: just want to know, what is remaining, what should take up on first priority
12:24:13 sfinucan johnthetubaguy: Looks like the work of dansmith (https://github.com/openstack/nova/commit/43063888)
12:24:33 gibi diga_: I think nothing was done on the notification side in the placement work
12:24:36 johnthetubaguy yeah, was just digging into that
12:24:58 gibi diga_: one I saw a WIP patch from cdent
12:25:05 gibi diga_: let me dig up a link
12:25:09 diga_ gibi: ok
12:25:12 johnthetubaguy sfinucan: I mean if you set everything to -1, you get the same behaviour as no-op I think
12:25:16 openstackgerrit Stephen Finucane proposed openstack/nova: conf: Deprecate yet another nova-net option https://review.openstack.org/416959
12:25:20 cdent diga_, gibi : https://review.openstack.org/#/c/308503/
12:25:24 cdent there's very little there
12:25:30 johnthetubaguy sfinucan: I totally approved the change, just looked odd at a second glance
12:26:04 diga_ cdent: yes, this is one you posted in the meeting
12:26:05 cdent diga_: feel free to take that over, or ignore it, whichever seems best
12:26:07 gibi cdent: thanks, that one :)
12:26:11 sfinucan johnthetubaguy: Ah, so we only have two "plugins" and one of them can do the work of the other? Definitely worth killing then
12:26:18 johnthetubaguy sfinucan: tempting to not move the driver one, given its needed rework
12:26:22 diga_ cdent: sure
12:26:38 johnthetubaguy sfinucan: well, one is much slower at doing nothing, I presume, but I don't know so
12:26:46 diga_ cdent: just want to know how to run placement-api seperately
12:26:48 gibi diga_: also jaypipes drafted the blueprint with some TODO list https://blueprints.launchpad.net/nova/+spec/placement-notifications
12:26:58 johnthetubaguy sfinucan: I suspect its only a test hang over thing
12:27:01 diga_ gibi: yes, he assigned to me
12:27:04 sfinucan johnthetubaguy: I can't see why someone would want to disable quotas entirely, tbh
12:27:06 sfinucan Yes, exactly
12:27:34 gibi diga_: based on that bp, the next thing is a small spec describing what notifications the placement service wants to emit
12:28:00 diga_ gibi: okay
12:28:13 gibi diga_: do you need a link for the spec template?
12:28:20 cdent diga_: What do you mean by "separately"? It is automatically enabled in devstack if nova is enabled and runs under mod_wsgi. Do you want to run it without devstack?
12:28:27 diga_ gibi: yeah, will be helpful
12:28:37 sfinucan Hmm, OK I hold off on a reworked version that doesn't move driver
12:28:48 diga_ cdent: I just see n-api in devstack
12:28:59 diga_ cdent: is that all ?
12:29:06 sfinucan I also have this to deprecate that floating IPs opt so perhaps we shouldn't move that too (but I don't care so much about this one) https://review.openstack.org/#/c/416959/
12:29:12 gibi diga_: here is the template you need to fill https://github.com/openstack/nova-specs/blob/master/specs/pike-template.rst
12:29:23 diga_ gibi: okay
12:29:58 cdent diga: the service is called placement-api and it will be enabled unless it is explicitly turned off
12:29:59 gibi diga_: then push the spec up for review in the nova-specs repo
12:30:01 diga_ gibi: will go through it first
12:30:28 diga_ okay
12:30:31 diga_ cdent: okay
12:31:43 diga_ gibi: will submit the spec soon
12:32:08 diga_ gibi: I think it is necessary to submit the spec spec & then start the implementation
12:32:11 diga_ ?
12:32:26 sfinucan mdbooth: +2'd
12:33:16 gibi diga_: you can start the implementation as early as you would like but you can only expect code review on your implementation after the related spec is approved
12:33:40 diga_ gibi: ohh..kk
12:33:59 gibi diga_: you don't have to rush as nova folks are pretty busy finishing Ocata these days
12:33:59 diga_ gibi: got it, will start on spec first then
12:34:28 gibi diga_: cool
12:34:33 diga_ gibi: yep
12:37:19 openstackgerrit Stephen Finucane proposed openstack/os-vif: vif_plug_ovs: Always set MTU when plugging devices https://review.openstack.org/408779
12:38:56 sfinucan johnthetubaguy: Ah, mriedem already pipped us to the mark https://review.openstack.org/#/c/410996/
12:39:35 gibi diga_: here are two example spec about adding new notifications https://review.openstack.org/#/c/321336/ and https://review.openstack.org/#/c/340715/
12:40:34 diga_ gibi: Thanks for sharing this, this is very helpful :)
12:41:43 diga_ gibi: will go through it, add it accordingly
12:41:47 diga_ gibi: brb
12:46:44 openstackgerrit Zhenyu Zheng proposed openstack/nova: Add query parameters white list for server list/detail https://review.openstack.org/408571
13:07:07 openstackgerrit Chris Dent proposed openstack/nova: [WIP] Structure for simply managing placement-api-ref https://review.openstack.org/409340
13:08:00 cdent sdague: if you get a chance to look at that ^ to sanity check that would be awesome. It's not done, but I want to make sure I'm not heading down a bad path. You'll also likely have some insight on how to make the testing it tries to do happen in the right place.
13:12:57 sdague cdent: looking
13:13:05 cdent s
13:13:05 cdent awesome, thank
13:16:56 johnthetubaguy sfinucan: ah, good spot
13:19:04 sdague cdent: yeh, that catch looks generally good, though the gate docs target actually doesn't run tox -e docs, so that's not going to fail in the gate when it's failing locally
13:19:14 cdent yeah
13:19:19 cdent (mentioned that in the commit message)
13:19:44 openstackgerrit Alex Szarka proposed openstack/nova: Transform instance-exists notification https://review.openstack.org/403660
13:20:01 cdent it not failing is fine for now since there aren't any docs, but it would be nice to flip a switch at the bottom of the stack after all existing routes are docced
13:20:26 cdent s/bottom/top/ <- dan's ordering
13:22:01 johnthetubaguy sfinucan: I think we just move them all
13:30:19 sdague cdent: well, it causes the local tox rule to fail
13:31:12 cdent sdague: yeah, what I mean is: it would be nice to make the lack of docs to cause a docs failure (somehow), after all the existing routes are documented
13:38:20 mdbooth sfinucan: Thanks.
13:40:52 openstackgerrit Gleb Stepanov proposed openstack/python-novaclient: Add evacuate host method https://review.openstack.org/349179
13:41:36 sfinucan johnthetubaguy: I agree
13:43:29 openstackgerrit Alex Szarka proposed openstack/nova: Transform instance.rebuild notification https://review.openstack.org/384922
14:00:07 wznoinsk frickler, we're on it, thanks
14:10:29 johnthetubaguy cdent: I got a bit confused with a bit of gabbi, can you check I am interpreting it correctly please: https://review.openstack.org/#/c/407742
14:10:37 johnthetubaguy ah, oops, he is not here
14:14:48 edleafe johnthetubaguy: I'm kinda fuzzy on that part, too
14:23:40 openstackgerrit Gleb Stepanov proposed openstack/nova: Add _rebuild_instance helper method https://review.openstack.org/376004
14:36:29 openstackgerrit Alex Szarka proposed openstack/nova: Transform instance-exists notification https://review.openstack.org/403660
14:39:42 macsz sfinucan: my gerrit weltschmerz-ism precedes me
14:46:42 openstackgerrit sahid proposed openstack/nova: libvirt: fix uncaugth exception during soft reboot https://review.openstack.org/376350
14:46:43 openstackgerrit sahid proposed openstack/nova: libvirt: make coherent logs when reboot success https://review.openstack.org/416994
14:46:43 openstackgerrit sahid proposed openstack/nova: libvirt: rework _destroy method https://review.openstack.org/416995
14:50:14 mriedem wznoinsk: intel nfv ci having some issues today?
14:52:12 openstackgerrit Jeremy Liu proposed openstack/nova: [2/3]Replace six.iteritems() with .items() https://review.openstack.org/413470
14:52:43 openstackgerrit Marco Voelz proposed openstack/nova-specs: Add port deletion on termination of server https://review.openstack.org/416997
14:53:47 voelzmo @johnthetubaguy remember our conversation about pre-created ports being deleted on VM termination? proposed that to nova-specs for pike: ^^
15:01:46 mriedem voelzmo: just left some comments
15:02:54 mriedem personally i'd consider that super low priority since if you can create the port outside of nova you should be able to delete it too
15:03:45 mriedem the delete_on_termination stuff with volumes has gotten us into problems with cinder state management too
15:03:56 cdent sdague: having two different api-ref's in the nova tree makes for some ick with trying to reuse existing api-ref jobs and job groups. Would you suggest making a standalone placement-api-ref job and annotating it appropriately? Or something else?

Earlier   Later