Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2018-10-10
21:27:17 mriedem because the API will let you do that
21:27:34 mriedem there is no rate limiting on multi-create requests
21:27:46 mriedem and that size of request will melt your scheduler
21:28:03 mriedem related: https://review.openstack.org/#/c/510235/
21:28:08 sorrison is the multi create number set in the request spec? I can have a look in the db to see what our stats are like
21:28:38 mriedem yes, it's the request spec "num_instances" field
21:35:39 openstackgerrit Merged openstack/os-vif master: clean up ip_command interface https://review.openstack.org/609414
21:36:32 sorrison mriedem: very tricky to get that info out of mysql due to json blob. We're not running a version of mysql that has json support sadly
21:36:51 mriedem damn
21:40:38 melwitt I never knew request spec was a json blob until now O.o
21:40:58 openstackgerrit Matt Riedemann proposed openstack/nova master: Don't log error in _remove_deleted_instances_allocations if compute is new https://review.openstack.org/609552
21:41:18 sorrison either way, we're aware of the issue but haven't had any major issues
21:41:42 mriedem maybe i can get some data from some public cloud ops
21:41:46 sorrison Just trying to get my sql foo on to see if I can extract the num_instances
21:51:46 sorrison ok so max num_instances we've had is 49
21:52:48 sorrison mriedem: stats here http://paste.openstack.org/show/731867/
21:53:23 melwitt neat
21:56:53 mriedem sorrison: nice, thanks
22:01:13 sorrison mriedem: I sorted out https://review.openstack.org/#/c/608474/ still not sure about correct name for policy as it does affect list and show for a flavor
22:01:55 mriedem that's not checked on show is it?
22:02:05 mriedem or you mean, allow non-admins to show private flavors that they don't have access to?
22:02:21 mriedem i.e. support person trying to triage a bug for a server created with a private flavor?
22:07:40 sorrison Yes
22:07:55 sorrison The change to allow policy for a flavor show is https://review.openstack.org/#/c/608474/3/nova/objects/flavor.py
22:08:10 sorrison it's for reporting scripts
22:09:16 mriedem yup took me a second to sort that out
22:09:31 sorrison yeah it goes down a few layers from the api
22:12:29 mriedem sorrison: ok comments inline;
22:12:41 mriedem i left some suggestions about the rule name, but they aren't awesome
22:12:49 mriedem maybe alex_xu or gmann or dansmith have ideas
22:15:06 sorrison yeah I can't think of something good there
22:16:04 sorrison with disabled flavors I think they are useful. We are planning on retiring some flavors soon and were planning on disabling them by updating the DB
22:16:27 sorrison unless there is a better way to retire flavors?
22:27:33 mriedem there is no way to disable flavors via the API, which is why i'm sort of hesitant to mention them
22:27:49 mriedem but it's not a big deal to leave that in if you're hacking flavors.disabled via the db directly
22:27:55 mriedem it just sucks you have to do that...
22:28:09 mriedem we have a PUT /flavors/{flavor_id} now...seems that would be a place to disable/enable flavors
22:28:16 melwitt AFAIK, people retired flavors by deleting them. and I remember we had a bug back from eons ago where 'nova show' would fail on existing instances with the retired flavor because it was trying to pull a flavor that was deleted from the db
22:28:43 melwitt that got fixed by flavors embedded on instances
22:28:57 mriedem and we expose the flavor details embedded in the instance in the API now
22:34:15 sorrison hmm yeah that makes sense and can prob just retire by deleting now
22:57:36 mriedem efried: see the thread on the ML about moving taskflow out of openstack governance?
22:57:40 mriedem powervm might care about that
22:58:54 mriedem oh heh i see you did :)
22:59:02 openstackgerrit Matt Riedemann proposed openstack/nova master: Skip _remove_deleted_instances_allocations if compute is new https://review.openstack.org/609552
23:00:14 openstackgerrit Matt Riedemann proposed openstack/nova master: Skip _remove_deleted_instances_allocations if compute is new https://review.openstack.org/609552
23:44:04 openstackgerrit iain MacDonnell proposed openstack/nova master: Handle online_data_migrations exceptions https://review.openstack.org/608091
#openstack-nova - 2018-10-11
00:05:22 mnaser mnaser
00:31:55 openstackgerrit Brin Zhang proposed openstack/python-novaclient stable/pike: Redirect the old release note url https://review.openstack.org/609566
01:52:21 openstackgerrit Merged openstack/nova master: api-ref: Remove a description in servers-actions.inc https://review.openstack.org/608796
01:52:29 openstackgerrit Merged openstack/nova master: Remove useless TODO section https://review.openstack.org/608802
03:03:39 openstackgerrit Merged openstack/nova master: remove commented-out code https://review.openstack.org/605635
03:06:41 openstackgerrit Artom Lifshitz proposed openstack/nova master: WIP: Handle volume API failure in post_live_migration https://review.openstack.org/609517
03:19:13 openstackgerrit Tao Li proposed openstack/nova master: Cleanup the instance when MessageDeliveryFailure exception https://review.openstack.org/608500
03:24:01 openstackgerrit Merged openstack/nova stable/rocky: Imported Translations from Zanata https://review.openstack.org/604260
05:53:56 openstackgerrit Sam Morrison proposed openstack/nova master: Allow ability for non admin users to list all flavors. https://review.openstack.org/608474
05:55:26 openstackgerrit Juan Antonio Osorio Robles proposed openstack/nova master: Remove saved_file_rules check from policy https://review.openstack.org/609591
06:37:10 openstackgerrit Naichuan Sun proposed openstack/nova master: xenapi(N-R-P):Get vgpu info from `allocations` https://review.openstack.org/521717
06:59:45 openstackgerrit Naichuan Sun proposed openstack/nova master: xenapi(N-R-P):Get vgpu info from `allocations` https://review.openstack.org/521717
07:00:00 openstackgerrit Naichuan Sun proposed openstack/nova master: xenapi(N-R-P): support compute node resource provider update https://review.openstack.org/521041
07:39:35 openstackgerrit Naichuan Sun proposed openstack/nova master: os-xenapi(n-rp): add traits for vgpu n-rp https://review.openstack.org/604269
08:13:54 openstackgerrit Juan Antonio Osorio Robles proposed openstack/nova master: Remove saved_file_rules check from policy https://review.openstack.org/609591
08:14:33 naichuans bauzas: Hi, Sylvain, any feedback about the vgpu-stein blueprint?
08:19:31 bauzas naichuans: I was trapped by some customer problem for the last 2 days, but you're in my pipe :)
08:23:31 naichuans bauzas: Got it, thank you very much :)
08:24:03 bauzas naichuans: I saw you also provided a reshape change
08:24:13 bauzas naichuans: I guess you looked at mine ?
08:24:43 naichuans bauzas: right, the first verson is completely same with yours~
08:24:56 bauzas naichuans: okay
08:25:13 bauzas naichuans: just to make it clear, I don't support multiple types in the same change
08:25:24 bauzas it's by decision
08:25:35 bauzas naichuans: people will need first to reshape
08:25:55 bauzas and then once we create a new change for having more than one, we need to make sure that the reshape is done
08:26:16 bauzas but that's a separate change I haven't written yet
08:26:21 naichuans bauzas: s/verson/version. Currently we pending the multiple vgpu type support, too. And the reshape works well on our test environment.
08:26:33 bauzas naichuans: I tested it too on my devstack box
08:28:28 naichuans bauzas: I'm also waiting the reshape works done, which is the reason we pending the multiple vgpu type support :)
08:28:50 bauzas k
08:36:19 openstackgerrit Yikun Jiang proposed openstack/nova-specs master: Support initial allocation ratios https://review.openstack.org/552105
09:07:25 BlackDex Hello there. i have an instance running with just a terminal. And i wonder how i can change the size of the console. Currently it has a small dimension (width&height) is there a way that i can change this?
09:16:38 openstackgerrit Balazs Gibizer proposed openstack/nova-specs master: Remove force flag from live-migrate and evacuate https://review.openstack.org/609330
09:30:25 openstackgerrit Merged openstack/python-novaclient master: doc: Start using openstackdoctheme's extlink extension https://review.openstack.org/608829
10:01:52 openstackgerrit Ghanshyam Mann proposed openstack/nova master: Merge used_limits extension response into limit view builder https://review.openstack.org/606031
10:46:25 s10 Hello. Should be bug https://bugs.launchpad.net/python-novaclient/+bug/1743532 be reopened because fix was reverted in https://review.openstack.org/#/c/572539 ?
10:46:25 openstack Launchpad bug 1743532 in python-novaclient "nova list doesn't work above api max_limit" [Undecided,Fix released] - Assigned to Jacek Tomasiak (skazi)
11:13:52 sean-k-mooney s10 from the message it looks like you could have passed --limie=-1 before so the change was not needed and intoduced a bug
11:17:07 lyarwood ~.
11:32:46 openstackgerrit Josephine Seifert proposed openstack/nova-specs master: Spec for the Nova part of Image Encryption https://review.openstack.org/608696
12:10:48 s10 sean-k-mooney: most of novaclient users don't expect, that they should pass limit=-1 to get all nova instances. For example, behaviour of the minderclient is expected - it gets all cinder volume. But for nova we should pass limit=-1, and nobody does it.
12:11:05 s10 sean-k-mooney: examples are https://github.com/openstack/ceilometer/blob/master/ceilometer/nova_client.py#L146 and https://github.com/openstack/watcher/blob/master/watcher/common/nova_helper.py#L76
12:11:37 sean-k-mooney s10: sure but the api limit is there to prevent excess load on that api and the operator does not expect that to be bypassed
12:11:51 s10 sean-k-mooney: ceilometer and watcher expects to get all nova instances here. If someone else uses novaclient python library, they also will face this issue
12:12:19 sean-k-mooney s10: ceilometer and watcher can easilly jsut pass -1 however
12:12:55 s10 sean-k-mooney: so basically we should open two bug in ceilometer and watcher and mark 1743532 as a feature, not a bug...
12:13:42 s10 sean-k-mooney: and should we open a bug against cinderclient? because it returns full list of volumes and don't honor api max_limit
12:14:29 sean-k-mooney on the cinder side proably unless its doing the paging internally.
12:14:55 sean-k-mooney so this is really a question of what are the expected semantics of the api and the client
12:15:35 sean-k-mooney i understand that users expect nova list to return all instance if invoked form the commandline
12:16:12 sean-k-mooney that is somwhat resonable but that could be supproted by defaulting to --limit=-1
12:18:19 sean-k-mooney s10: to be clear i personally also expect nova list to return all instances but the previous fix was reverted because it actully intoduced a bug so if we want alter teh behavior again it shoudl likely be done as an RFE

Earlier   Later