| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2021-01-21 | |||
| 19:08:04 | openstackgerrit | Merged openstack/nova master: zuul: Skip test_attach_cloned_encrypted_volume in nova-ceph-multistore https://review.opendev.org/c/openstack/nova/+/771777 | |
| 19:34:05 | openstackgerrit | Lee Yarwood proposed openstack/nova stable/queens: libvirt: Do not reference VIR_ERR_DEVICE_MISSING when libvirt is < v4.1.0 https://review.opendev.org/c/openstack/nova/+/747362 | |
| 19:35:10 | lyarwood | melwitt / elod ; https://review.opendev.org/c/openstack/nova/+/747362/ would you mind looking at this stack of stable/queens changes again, should be good to go now. | |
| 19:36:06 | lyarwood | I just tried to rebase that in the new gerrit UI and it has ended up with a merge conflict, wat?! | |
| 19:38:57 | lyarwood | ha, it just didn't report an actual conflict during the rebase attempt, fun | |
| 19:40:12 | openstackgerrit | Lee Yarwood proposed openstack/nova stable/queens: sync_guest_time: use the proper errno https://review.opendev.org/c/openstack/nova/+/742423 | |
| 19:40:13 | openstackgerrit | Lee Yarwood proposed openstack/nova stable/queens: libvirt: Remove reference to transient domain when detaching devices https://review.opendev.org/c/openstack/nova/+/608186 | |
| 19:40:14 | openstackgerrit | Lee Yarwood proposed openstack/nova stable/queens: libvirt: Handle VIR_ERR_DEVICE_MISSING when detaching devices https://review.opendev.org/c/openstack/nova/+/742424 | |
| 19:40:15 | openstackgerrit | Lee Yarwood proposed openstack/nova stable/queens: libvirt: Do not reference VIR_ERR_DEVICE_MISSING when libvirt is < v4.1.0 https://review.opendev.org/c/openstack/nova/+/747362 | |
| 19:40:36 | lyarwood | there rebased and hopefully ready | |
| 20:02:39 | elod | lyarwood: ack, I will review tomorrow :) | |
| 20:12:37 | openstackgerrit | Lee Yarwood proposed openstack/nova master: docs: Add reference docs for internal block device structures https://review.opendev.org/c/openstack/nova/+/771711 | |
| 20:34:05 | melwitt | sean-k-mooney: I read the backscroll from the nova meeting but I don't understand how it answers the question I had on the review, how can an operator prevent a user from leveraging numa aware vswitch? | |
| 20:34:51 | melwitt | "As an operator i wish to utilize numa aware vswitch but still be able to disable it for individual vm interfaces." | |
| 20:40:01 | sean-k-mooney | melwitt: what i ment by that was and still allow the tenant to disable it for indivigual vm interfaces | |
| 20:40:20 | sean-k-mooney | i had a other part of the spec that i planned to do but have not included too | |
| 20:40:37 | sean-k-mooney | which was to add a config option for the default polcy for numa vswitches | |
| 20:41:08 | melwitt | sean-k-mooney: ok, so tenant instead of operator. and that means there's a numa_affinity_policy that means disable? or by disable do you mean opt out | |
| 20:41:19 | sean-k-mooney | line 179 https://etherpad.opendev.org/p/nova-wallaby-ptg | |
| 20:41:55 | sean-k-mooney | melwitt: so if you have numa vsiwtchs configured thne you get strict affinity for every numa vm on the host | |
| 20:42:12 | sean-k-mooney | withthe port policy you coudl set it to prefer | |
| 20:42:25 | sean-k-mooney | which is basically best effort or no affinity enforced | |
| 20:42:53 | melwitt | ok, so there's not a concept of "disabling" it i.e. preventing it | |
| 20:43:07 | sean-k-mooney | the prefer policy is the closet thing | |
| 20:43:16 | sean-k-mooney | there is not way to request anti affinity | |
| 20:43:26 | sean-k-mooney | or disable all numa checks | |
| 20:43:34 | melwitt | ok, thanks | |
| 20:43:43 | sean-k-mooney | you can just say for this 1 interface do X | |
| 21:21:17 | openstackgerrit | Merged openstack/nova-specs master: add port scoped numa affinity spec https://review.opendev.org/c/openstack/nova-specs/+/765901 | |
| #openstack-nova - 2021-01-22 | |||
| 00:02:44 | brinzhang | sean-k-mooney, gibi: got it, thanks, will check | |
| 00:03:45 | gmann | lbragstad: left comment for placement policy - https://review.opendev.org/c/openstack/placement/+/760240/14/placement/tests/functional/fixtures/gabbits.py#763 | |
| 00:04:27 | gmann | lbragstad: if wee can run same or new tests with fixture to check if old token still work or not when enforce_scopy is false | |
| 00:06:23 | gmann | which is what we ship as default. | |
| 00:17:40 | brinzhang | sean-k-mooney: the os-simple-tenant-usage action will replace with os-simple-project-usage action, the tempest check the os-simple-tenant-usage action, if I dont skip the tempest check, it seems cannot pass | |
| 00:17:48 | brinzhang | sean-k-mooney: is it right? | |
| 00:18:07 | brinzhang | the nova change: https://review.opendev.org/c/openstack/nova/+/768852/7/nova/api/openstack/compute/routes.py#689 | |
| 00:20:03 | brinzhang | gmann: I know you are an expert in tempest, can you help to check? | |
| 00:20:30 | brinzhang | gmann: the tempest patch https://review.opendev.org/c/openstack/tempest/+/769120 | |
| 00:22:01 | gmann | brinzhang: sure, you want to move to openstack-qa ? | |
| 00:24:07 | brinzhang | gmann: of course | |
| 00:50:48 | gmann | brinzhang_: replied on gerrit for your query about os-simple-tenant-usage action https://review.opendev.org/c/openstack/nova/+/768852 | |
| 00:51:19 | gmann | brinzhang_: we still nee old API route and controller/sample/test etc as this is still valid for old microversion | |
| 00:53:06 | gmann | we do not have microversion knowledge in routes so we can add new one with old one still there and in API method in controller we can return 404 for old API for >2.89 | |
| 00:53:44 | brinzhang_ | gmann: greate, will check | |
| 00:54:02 | gmann | which is nothing but this @wsgi.Controller.api_version("2.1", "2.89") | |
| 00:55:09 | gmann | and new API method can be decorated with @wsgi.Controller.api_version("2.90") | |
| 00:55:28 | gmann | so any user requesting new API will get 404 with <2.90 | |
| 00:57:14 | gmann | basically keep the old API things (test/sample etc) as it is but with capping it with microversion 2.89. | |
| 00:58:46 | brinzhang_ | you mean, I should add @wsgi.Controller.api_version("2.90") in the index and show api? | |
| 00:59:02 | brinzhang_ | s/in/on/ | |
| 00:59:26 | gmann | for new API yes | |
| 00:59:50 | brinzhang_ | https://review.opendev.org/c/openstack/nova/+/768509/10/nova/api/openstack/compute/simple_tenant_usage.py | |
| 01:00:29 | brinzhang_ | gmann: I think I should add two api for 2.90, and add the max version limit the old version | |
| 01:00:56 | openstackgerrit | Lance Bragstad proposed openstack/placement master: Implement secure RBAC for resource providers https://review.opendev.org/c/openstack/placement/+/760240 | |
| 01:00:58 | gmann | like ServerTopology API added in 2.78 https://github.com/openstack/nova/blob/master/nova/api/openstack/compute/server_topology.py#L26 | |
| 01:01:03 | brinzhang_ | gmann: I am not understand waht you say about the router change | |
| 01:01:38 | gmann | brinzhang_: in routes, you just need to add the new controller and url path and leave old one as it is | |
| 01:02:09 | brinzhang_ | ack | |
| 01:02:31 | gmann | because old API still valid for older microversion. | |
| 01:02:41 | brinzhang_ | gmann: do we need to rename the tenant file? | |
| 01:02:50 | gmann | which one? | |
| 01:03:24 | gmann | this one? https://github.com/openstack/nova/blob/master/nova/api/openstack/compute/simple_tenant_usage.py | |
| 01:03:29 | brinzhang_ | this patch I add the 2.90 support https://review.opendev.org/c/openstack/nova/+/768509/10/nova/api/openstack/compute/simple_tenant_usage.py | |
| 01:04:56 | brinzhang_ | and https://review.opendev.org/c/openstack/nova/+/768852 renamed the tenant* to project* | |
| 01:06:06 | gmann | brinzhang_: there are two part 1. changing the 'tenant' in response etc which is 768509 this is good as you are doing, I had a quick glance | |
| 01:06:45 | brinzhang_ | https://review.opendev.org/c/openstack/nova/+/768509/10/nova/api/openstack/compute/simple_tenant_usage.py#295 | |
| 01:07:10 | gmann | and 2nd is about adding new API method which is 768852 for this we can add new file or new method using old method in existing file | |
| 01:07:42 | brinzhang_ | this patch I should add the @wsgi.Controller.api_version("2.1", "2.89") in the old index and show api, and add @wsgi.Controller.api_version("2.90") to as the new api for index and show, right? | |
| 01:07:58 | gmann | right | |
| 01:08:36 | brinzhang_ | but this change in the 768852 | |
| 01:08:48 | brinzhang_ | as your suggestion | |
| 01:08:58 | gmann | have two option here 1. if rename file then add new controller class which can reuse the old APi method 2. have complete new file with new method which can also use existing method | |
| 01:09:40 | gmann | may be 1st option is better. you can rename file and keep old controller class in that file with all API method as @wsgi.Controller.api_version("2.1", "2.89") | |
| 01:10:01 | gmann | and new controller class with new method with @wsgi.Controller.api_version("2.90") and add these in routes.py | |
| 01:12:25 | gmann | and same way in tests also we can do the same to reuse the code | |
| 01:13:13 | brinzhang_ | gmann: yes, in 768852, we renamed the file, and reused the old test | |
| 01:14:31 | brinzhang_ | I need to keep the old action in the router and add the new action controller | |
| 01:14:50 | gmann | brinzhang_: but you are changing old API route and controller which means old API is completely gone - https://review.opendev.org/c/openstack/nova/+/768852/7/nova/api/openstack/compute/simple_project_usage.py | |
| 01:15:48 | brinzhang_ | is it need to restort the nova/api/openstack/compute/simple_tenant_usage.py | |
| 01:15:49 | brinzhang_ | file? | |
| 01:15:52 | gmann | for example: GET /os-simple-tenant-usage wil return 404 even for older microversion | |
| 01:15:59 | brinzhang_ | and add the nova/api/openstack/compute/simple_project_usage.py? | |
| 01:16:15 | gmann | that is one option | |
| 01:16:18 | gmann | or other option is | |
| 01:17:22 | gmann | rename and add SimpleProjectUsageController(SimpleTenantUsageController) as new class there | |
| 01:18:25 | gmann | 'GET': [simple_project_usage_controller, 'index'] | |
| 01:18:25 | gmann | and in routes.py you can add new path like ('/os-simple-project-usage', { | |
| 01:19:04 | gmann | def index() in new controller will just call the old SimpleTenantUsageController.index() and have @wsgi.Controller.api_version("2.90") | |
| 01:19:49 | brinzhang_ | gmann: ok, we will try to use the two option | |
| 01:20:05 | brinzhang_ | a question | |
| 01:21:00 | brinzhang_ | @wsgi.Controller.api_version("2.1", "2.89") and @wsgi.Controller.api_version("2.90") for the api change, is it need to change with the router change? | |
| 01:21:15 | brinzhang_ | I mean keep this change in 768852 patch | |
| 01:21:57 | brinzhang_ | gmann: what do you think? | |
| 01:25:29 | gmann | brinzhang_: in routes.py if you rename the file then 1. for old API route, change import (from nova.api.openstack.compute import simple_tenant_usage -> from nova.api.openstack.compute import simple_project_usage ) | |
| 01:25:47 | gmann | and for new API add new route as separate | |
| 01:29:07 | gmann | brinzhang_: like this http://paste.openstack.org/show/801854/ | |
| 01:30:26 | brinzhang_ | gmann: yeah, get it | |
| 01:31:28 | brinzhang_ | gmann: above question is which patch I shuold change the max or min version changing of the index and show usages apis? | |
| 01:32:06 | gmann | but other file like request/response sample file can exist in separate if it become very lengthy and unreadable | |
| 01:32:53 | gmann | brinzhang_: ah that is good question but tricky also. | |