| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2017-03-22 | |||
| 12:44:13 | alex_xu | cdent: \o/ | |
| 12:45:16 | alex_xu | nova api meeting is in 15 mins at #openstack-meeting-4 | |
| 12:47:40 | openstackgerrit | Takashi NATSUME proposed openstack/python-novaclient master: Microversion 2.42 - Fix tag attribute disappearing https://review.openstack.org/429512 | |
| 12:47:58 | openstackgerrit | Takashi NATSUME proposed openstack/python-novaclient master: Microversion 2.43 - Enable cold migration with target host https://review.openstack.org/406707 | |
| 12:48:08 | openstackgerrit | Takashi NATSUME proposed openstack/python-novaclient master: Microversion 2.43: List/Show all server migration types https://review.openstack.org/430839 | |
| 12:52:19 | openstackgerrit | Takashi NATSUME proposed openstack/nova master: api-ref: Fix parameters in os-hosts.inc https://review.openstack.org/416444 | |
| 12:54:09 | openstackgerrit | Takashi NATSUME proposed openstack/nova master: Avoid forcing translation on logging calls https://review.openstack.org/413876 | |
| 12:59:36 | jgriffith | lyarwood yep, thanks! I'll get that refactored with your patches now and back up | |
| 13:06:59 | sfinucan | mriedem, johnthetubaguy: Could you take a look at [1] again? Think I've all your feedback addressed [1] https://review.openstack.org/#/c/364468/ | |
| 13:09:40 | mriedem | sfinucan: yeah, and jaypipes | |
| 13:09:43 | mriedem | after api meeting | |
| 13:09:51 | sfinucan | (y) | |
| 13:19:33 | lyarwood | jgriffith: cool thanks | |
| 13:23:17 | mriedem | dansmith: looks like in the services/hypervisors uuids spec, i missed the os-hosts api https://review.openstack.org/#/c/447149/ which is basically the same as the os-services api, except some other actions can be performed like stop/start/reboot of a host | |
| 13:23:36 | mriedem | but i'm guessing when you made os-services tolerate multiple cells, we got os-hosts support for free | |
| 13:23:41 | mriedem | since they work on the same resource | |
| 13:24:28 | jaypipes | mriedem: unfortunately. | |
| 13:25:47 | mriedem | jaypipes: right :) | |
| 13:25:52 | mriedem | makes me want to merge the APIs | |
| 13:26:01 | jaypipes | mriedem: err, no | |
| 13:26:08 | mriedem | err no? | |
| 13:26:12 | jaypipes | mriedem: compute node != the RPC service worker. | |
| 13:26:17 | mriedem | this isn't that | |
| 13:26:21 | mriedem | os-hypervisors == compute nodes | |
| 13:26:24 | jaypipes | one is a communication target, the other is a resource. | |
| 13:26:25 | mriedem | os-hosts == services | |
| 13:26:28 | mriedem | os-services == services | |
| 13:26:29 | jaypipes | oh... | |
| 13:26:35 | mriedem | i'm talking about the resource representation in those APIs | |
| 13:26:36 | jaypipes | shit, sorry Matt. | |
| 13:26:41 | mriedem | os-hosts and os-services are both services resources | |
| 13:26:50 | mriedem | both can list and show details, and enable/disable | |
| 13:26:50 | jaypipes | hmm. | |
| 13:26:59 | mriedem | but one has some more actions and the other has some other actions | |
| 13:27:03 | mriedem | but separate dumb APIs | |
| 13:27:16 | jaypipes | interesting. lemme finish review on sfinucan's PCI NUMA reserve bp and then will look at it. | |
| 13:27:40 | dansmith | mriedem: yeah, I dunno how we ended up with those | |
| 13:27:57 | mriedem | extensions? | |
| 13:28:01 | mriedem | i blame extensions | |
| 13:28:03 | jaypipes | mb | |
| 13:35:37 | dansmith | sfinucan: hmm, it didn't look like a random failure, but okay | |
| 13:36:16 | sfinucan | All I know is it worked on my machine 🙃 | |
| 13:39:13 | openstackgerrit | Vladik Romanovsky proposed openstack/nova master: neutron: retrieve physical network name from a multi-provider network https://review.openstack.org/430026 | |
| 13:39:24 | dansmith | sfinucan: this one seems like it might should have a reno about removing the thing, especially since it's firewall/security related no? https://review.openstack.org/#/c/430912/8 | |
| 13:40:04 | sfinucan | dansmith: Actually, yeah - I promised that at the PTG. I'll whip one up now | |
| 13:40:13 | dansmith | thanks | |
| 13:40:31 | sfinucan | fwiw, I posted a question on "who uses this" to the operators-list and got zilch back | |
| 13:46:33 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Remove aggregate uuid generation on load from DB https://review.openstack.org/448579 | |
| 13:46:48 | bauzas | mriedem: had a discussion with ttx this FR morning | |
| 13:46:54 | bauzas | mriedem: about the Forum | |
| 13:47:02 | bauzas | mriedem: not sure you saw that | |
| 13:47:12 | dansmith | sfinucan: cool | |
| 13:47:28 | bauzas | tl,dr: I just added a new section for our nova etherpad | |
| 13:47:37 | bauzas | mriedem: about things we could be discussing ad-hoc | |
| 13:48:12 | sfinucan | or not - looks like it got dropped :/ I'll add the reno then put that in -W for a week or so | |
| 13:48:46 | openstackgerrit | Danfly proposed openstack/nova master: remove mox from unit/api/openstack/compute/test_flavor_manage.py https://review.openstack.org/448581 | |
| 13:50:54 | bauzas | dansmith: jroll: added a new comment for explaining why I think we could have a possible problem for rolling-upgrade with https://review.openstack.org/#/c/446570/2 | |
| 13:51:09 | dansmith | I'm reading now | |
| 13:52:42 | jaypipes | sfinucan, mriedem: k, reviewed. | |
| 13:53:24 | sfinucan | jaypipes: Excellent. I'll take a look shortly | |
| 13:53:29 | sfinucan | Cheers :) | |
| 13:54:19 | openstackgerrit | Sujitha proposed openstack/nova master: Add description to policies in cells_scheduler.py https://review.openstack.org/448260 | |
| 13:55:19 | jroll | bauzas: replied | |
| 13:56:38 | bauzas | jroll: I thought we were asking operators to modify their flavors *before* upgrading to Pike | |
| 13:56:50 | bauzas | at least that's what we discussed at the PTG IIRC | |
| 13:57:43 | jroll | bauzas: right, and now we've found the catch and cannot do that | |
| 13:57:44 | openstackgerrit | dane-fichter proposed openstack/nova master: Use cursive for signature verification https://review.openstack.org/351232 | |
| 13:58:08 | jroll | bauzas: we were hoping to drop cpu/memory/disk from the ironic resource in pike, it'll need to happen in queens now | |
| 13:58:42 | bauzas | jroll: okay, then I think we could be adding that into the spec by the operator impact section | |
| 14:00:52 | jroll | bauzas: sure, I could add "after fully deploying pike" there if it makes you happy | |
| 14:01:44 | bauzas | jroll: waiting for dansmith's opinion :) | |
| 14:02:14 | dansmith | bauzas: my opinion has already been expressed as a +2 | |
| 14:02:44 | dane-fichter | mriedem: new patch shows how cursive integration would work with the next cursive release https://review.openstack.org/#/c/351232/12 | |
| 14:03:02 | mriedem | ok | |
| 14:03:24 | dane-fichter | mriedem: basically just making the key types a public member and accessing it directly in fields.py | |
| 14:03:51 | openstackgerrit | Stephen Finucane proposed openstack/nova master: conf: Final cleanups in conf/network https://review.openstack.org/431023 | |
| 14:03:51 | openstackgerrit | Stephen Finucane proposed openstack/nova master: conf: Deprecate 'allow_same_net_traffic' https://review.openstack.org/430913 | |
| 14:03:52 | openstackgerrit | Stephen Finucane proposed openstack/nova master: libvirt: Ignore 'allow_same_net_traffic' for port filters https://review.openstack.org/430912 | |
| 14:04:08 | mriedem | johnthetubaguy: looking at https://developer.openstack.org/api-ref/compute/#hosts-os-hosts it links to the hypervisor support matrix which doesn't mention anything about host power actions for stop/start/reboot :) it does mention the host maintenance mode thing though https://docs.openstack.org/developer/nova/support-matrix.html#operation_maintenance_mode and that it's only implemented for xen | |
| 14:04:40 | johnthetubaguy | mriedem: yeah, it doesn't do anything but change an icon in the management console | |
| 14:04:52 | mriedem | johnthetubaguy: i thought it auto-evacuated all guests from that xen host? | |
| 14:05:02 | mriedem | or something crazy like that | |
| 14:05:12 | mriedem | which i'd think would mess up resource tracking | |
| 14:05:29 | mriedem | like, i thought that was one of the primary reasons we said no to the hyperv cluster driver | |
| 14:05:36 | openstackgerrit | Andrey Volkov proposed openstack/nova-specs master: User-controlled SR-IOV ports allocation https://review.openstack.org/182242 | |
| 14:05:41 | johnthetubaguy | mriedem: no, thats just want people think it does | |
| 14:05:55 | mriedem | because it would auto load balance in the background cluster and we wouldn't know about the resource tracking changes, except the virt driver would hook back into the resource tracker and fiddle things for us, which we didn't want | |
| 14:06:08 | mriedem | johnthetubaguy: ok so it literally is only implemented by one virt driver, and does nothing :) | |
| 14:06:12 | johnthetubaguy | mriedem: yeah, that | |
| 14:06:18 | mriedem | that's aweoms | |
| 14:06:21 | mriedem | *awesome | |
| 14:06:23 | johnthetubaguy | agreed | |
| 14:06:46 | mriedem | i guess the description on https://docs.openstack.org/developer/nova/support-matrix.html#operation_maintenance_mode is a lie then | |
| 14:08:26 | dansmith | mriedem: yeah it's always been xen-only | |
| 14:08:38 | dansmith | and very specific to how its own maintenance mode works, IIRC | |
| 14:08:53 | dansmith | which is why we didn't use it for force-down | |
| 14:09:04 | johnthetubaguy | mriedem: yeah, I think it might stop new instances being built, but thats it | |
| 14:09:38 | johnthetubaguy | I think its time to kill os-hosts | |
| 14:09:45 | johnthetubaguy | in the deprecate sense of kill | |
| 14:10:03 | mriedem | johnthetubaguy: yeah, i'll send something to the ML to see if anyone is actually using it | |