Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2017-06-01
19:13:22 jaypipes mriedem: however, claims in the scheduler make things quite a bit simpler, which is why I've been waiting for that to get through
19:23:24 openstackgerrit Artom Lifshitz proposed openstack/nova master: Convert instance to primitive for metadata cache https://review.openstack.org/469978
19:28:25 melwitt mriedem: I think that should be all fine because the used limits API calls the same functions that are being switched over to counting. so it will just count server groups to show how much is used, it won't be using the usage records anymore
19:28:34 melwitt but I will double check, of course
19:31:21 mriedem melwitt: ok left some questions in here about the race thing too https://review.openstack.org/#/c/446240/14
19:31:25 jaypipes melwitt: you see my comment above? :)
19:31:27 mriedem i'm not sure i'm following that
19:31:42 melwitt mriedem: okay, will check and reply
19:33:59 mriedem avolkov also had comments/questions in PS13
19:34:57 melwitt jaypipes: ah, yes. hmm ... maybe the placement stuff actually, want to make sure that's on the right track from the scheduler experts
19:35:35 jaypipes melwitt: you got it.
19:35:43 melwitt thanks!
19:37:44 melwitt thanks, missed those comments from avolkov
19:40:37 openstackgerrit Matt Riedemann proposed openstack/nova-specs master: Fix to return server groups quota in the quota-classes API https://review.openstack.org/468619
19:51:22 openstackgerrit Jay Pipes proposed openstack/nova master: Sync os-traits to Traits database https://review.openstack.org/469578
19:55:53 openstackgerrit Artom Lifshitz proposed openstack/python-novaclient master: Better handle key=value pair not being key=value https://review.openstack.org/469988
20:04:52 openstackgerrit Artom Lifshitz proposed openstack/python-novaclient master: Better handle key=value pair not being key=value https://review.openstack.org/469988
20:13:30 edmondsw what conf setting defines the "default periodic interval"?
20:14:30 edmondsw lots of conf settings have help that says if they are left at 0 (default) then the "default periodic interval" is used, but give no clue as to where that is found
20:14:50 melwitt mriedem: replied to all the comments
20:14:56 mriedem edmondsw: it's hard-coded in oslo
20:15:20 mriedem https://github.com/openstack/oslo.service/blob/master/oslo_service/periodic_task.py#L29
20:15:24 openstackgerrit Artom Lifshitz proposed openstack/python-novaclient master: Error out if nic auto or none are not alone https://review.openstack.org/470000
20:15:38 artom There, a couple of paper cuts taken care of
20:15:53 edmondsw mriedem so 60? https://docs.openstack.org/developer/oslo.service/api/periodic_task.html#oslo_service.periodic_task.periodic_task
20:16:12 mriedem edmondsw: yes
20:16:15 edmondsw tx
20:16:39 mriedem artom: we should be able to tag a nic with auto
20:16:57 mriedem auto in most cases is the exact same behavior as not specifying anything for networks before 2.37
20:17:05 artom mriedem, eh? I though the API was explicit in now allowing anything else with 'auto'
20:17:16 artom *not
20:17:26 artom I mean, we *should*, sure
20:17:35 artom But the way it's currently defined, we can't
20:17:59 mriedem please hand me that knife
20:18:45 artom mriedem, https://github.com/openstack/nova/blob/master/nova/api/openstack/compute/schemas/servers.py#L88-L105
20:19:00 artom It's 'oneOf' - either auto/none, or a "real" nic definition
20:19:22 mriedem yeah, ffs
20:19:45 mriedem probably why i originally had it written for the schema to be auto/none for the uuid field in the dict
20:20:02 mriedem but it was suggested that rather than have networks={'uuid': 'auto'} we just have networks='auto'
20:20:29 artom In retrospect that wasn't smart
20:20:36 artom But I agree the first option is... weird
20:20:51 mriedem well, it also wasn't caught in any way as we knew because of 2.42
20:21:03 artom That didn't help, indeed
20:22:17 mriedem i suppose the api-ref should be updated, "The auto and none values cannot be used with any other network values, including other network uuids, ports or fixed IPs." to include "tag"
20:22:47 mriedem terrible terrible api
20:23:06 mriedem i guess you can pre-create the port in neutron, provide that to nova with the tag you want if you're an advanced user
20:23:21 mriedem but people don't even like doing that with volume types and boot from volume
20:24:20 cfriesen_ mriedem: doesn't make sense to me...if you script it it's no big deal
20:25:10 cfriesen_ besides, if you only have one vif from "networks='auto'" why would you need to tag it?
20:25:22 mriedem idk
20:25:24 mriedem maybe you don't
20:25:34 mriedem maybe we consider tags an advanced use case
20:25:40 mriedem and that's how we can justify now feeling bad about it
20:25:44 mriedem *not
20:25:58 mriedem whatever i can do to get through this day
20:29:14 cfriesen_ I see in Newton's online_data_migrations we go back and create a RequestSpec object for old instances. Is there stuff that we're going to end up missing since the info isn't available any more?
20:31:13 mriedem cfriesen_: that's a bauzas question
20:31:53 cfriesen_ okay, will ask him.
20:33:12 openstackgerrit Matt Riedemann proposed openstack/nova master: Update server create networks API reference description for tags https://review.openstack.org/470004
20:33:13 mriedem artom: ^
20:34:08 mriedem artom: maybe it's not so bad, could you even do networks={'tag': 'foo'} < 2.37?
20:34:22 mriedem i think it would puke if you specified networks w/o a network uuid or a port
20:35:40 mriedem yeah https://github.com/openstack/nova/blob/master/nova/api/openstack/compute/servers.py#L418
20:35:47 mriedem if port isn't specified, uuid is quired
20:35:49 mriedem *required
20:35:58 mriedem so you couldn't just provide a tag w/o anything else before
20:35:59 mriedem whew
20:37:36 artom mriedem, yeah, I don't think you could ever have a nic with *only* a tag
20:46:22 artom Oh hey, I got 470000!
20:47:10 mriedem is that a special number?
20:47:17 artom It's round
20:47:22 artom That's beer, right?
20:47:31 mriedem molson?
20:47:33 mriedem or labat?
20:47:40 artom Nah, I drink the fancy microbrews
20:48:01 mriedem so, michelob light ultra?
20:48:28 artom Don't have those here, but if I'm ever in... wherever they're from, I'll try
20:49:33 mriedem they're in 'merica
20:49:35 artom Lately I've taken a fancy to Vrooden's bitter
20:50:03 artom They're so hipster they don't even have their own domain: https://vrooden.wixsite.com/vrooden
20:50:18 artom 'merica eh? I hear it's dangerous for strangers there ;)
20:50:29 mriedem https://www.youtube.com/watch?v=M-MWN-XUtEQ is for when you're working hard all day being white and privileged, you need to relax with a bottle of water
20:51:09 mriedem they skipped the escalator
20:51:12 mriedem because they live in today
20:53:54 artom Dammit, http://www.checkmyprivilege.com/ is gone
21:02:02 openstackgerrit Artom Lifshitz proposed openstack/python-novaclient master: Error out if nic auto or none are not alone https://review.openstack.org/470000
21:32:04 mriedem cfriesen_: never fear, i'm going through your instance flavors api change
21:48:42 mriedem dansmith: cfriesen_: dhellmann fixed the missing renos issue https://review.openstack.org/#/c/470023/
21:49:15 mriedem just in time for everyone and their dog to be moving off mitaka
21:49:23 mriedem except penick
21:57:30 mriedem cfriesen_: reviewed
21:57:32 mriedem in great detail
22:03:16 cfriesen_ mriedem: thanks, will go check it out
22:03:56 cfriesen_ geez, could you add more stuff?
22:06:50 mriedem i don't think so
22:12:42 cfriesen_ mriedem: for the one that will be max_version 2.46, do you want it named as flavor_id_2_46?
22:14:33 mriedem looking
22:15:00 cfriesen_ and for "flavor_links" which is unversioned, did you want me to add a _2_46 version for clarity?
22:15:35 mriedem so flavor_id_body_2 -> flavor_id_body_2_46?
22:15:38 cfriesen_ that way all the sub-fields of flavor would be versioned

Earlier   Later