Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2017-10-25
09:06:48 gibi johnthetubaguy: sorry I have no idea when we delete cell_mappings
09:07:10 johnthetubaguy jmccarthy: not sure if this helps, but this is part of the create instance code: https://github.com/openstack/nova/blob/master/nova/virt/libvirt/driver.py#L2806
09:07:22 gibi johnthetubaguy: sitting on a meeting right now, I can try to look at the code after that
09:07:40 johnthetubaguy gibi: thanks, I think it might be never, which seems like a poor choice
09:16:17 johnthetubaguy gibi: I just remembered about doing a GET on a deleted instance, for admins
09:17:16 johnthetubaguy gibi: but we don't have an archive deleted rows thing for the API DB...
09:17:29 johnthetubaguy (as there is no soft delete)
09:26:54 jmccarthy johnthetubaguy: Thanks !
09:48:56 openstackgerrit Yikun Jiang proposed openstack/nova master: [WIP] Add migration db and object pagination support. https://review.openstack.org/514904
10:04:05 openstackgerrit sahid proposed openstack/nova master: libvirt: disconnect volume from host during detach https://review.openstack.org/515008
10:09:27 gibi johnthetubaguy: do you refert to the InstanceMapping class when you say cell_mappings?
10:10:03 johnthetubaguy gibi: sorry, I meant instance_mappings
10:13:51 gibi johnthetubaguy: I found some cleanup here https://github.com/openstack/nova/blob/df9fbbfec97a1ce13364d025bcd66621725bb7e4/nova/compute/api.py#L1200
10:14:29 gibi johnthetubaguy: but I guess this is just the case when nova fails to build the instance
10:15:15 johnthetubaguy gibi: looks that way
10:15:37 johnthetubaguy gibi: I suspect its because you can GET for a delete instance, so you can only delete once you do a hard delete in the child cell
10:16:04 johnthetubaguy gibi: thanks for looking into that for me though, backs up what I was thinking
10:19:45 gibi johnthetubaguy: as I see there is no way to soft delete an InstanceMapping in the api db so your observation makes sense
10:20:34 johnthetubaguy yeah, no soft delete in any of the api_db, I think
10:21:05 gibi johnthetubaguy: does the archiving process moves the instance mapping from the production table to the shadow table?
10:21:12 gibi johnthetubaguy: that could be a way to clean up
10:22:34 johnthetubaguy but that util doesn't have access to the api_db, generally
10:23:50 johnthetubaguy oh, wait, I was thinking about the instance moving to a shadow table, I didn't think we did shadow tables in the API db
10:24:13 johnthetubaguy I saw we had a non-shadow table archive, but I must have got mixed up
10:24:28 johnthetubaguy s/saw/swore/
10:26:42 gibi johnthetubaguy: OK then I conclude that nova doesn't delete the instance mapping at all
10:27:08 johnthetubaguy gibi: thanks, I thought I was going a bit mad not being able to track that down
10:37:27 masber good afternoon, I am having an issue with nova where my instances are failing during building process
10:39:40 masber this is the error I can see running openstack server show command http://paste.openstack.org/raw/624579/
10:46:43 johnthetubaguy masber: if you look at the original log, it has a request-id in there, you will want to search for other logs with that id in it, that will show you why you hit that error
10:54:04 masber johnthetubaguy, is this what you mean? http://paste.openstack.org/show/624582/
11:05:10 johnthetubaguy masber: maybe, but I suspect you need logs from nova-conductor and other services, using that request-id (although I don't see the no valid host error log in there)
11:05:35 masber oh ok
11:05:40 masber I just checked the scheduler
11:05:53 johnthetubaguy masber: should really move to the #openstack channel I guess, its more operator focused than dev
11:05:53 masber let me see other log files
11:09:08 masber johnthetubaguy, ok I got a permission denied http://paste.openstack.org/raw/624585/
11:09:53 johnthetubaguy masber: ah, cool, that sounds like the root cause
11:10:48 masber how weird
11:11:52 ttsiouts hey guys! I started investigating this bug report: https://bugs.launchpad.net/nova/+bug/1644457. was there a decision that a user's keypairs in_use should not be counted by the cli: nova quota-show --user --detail?
11:11:53 openstack Launchpad bug 1644457 in OpenStack Compute (nova) "keypair quota error" [Medium,Confirmed] - Assigned to Theodoros Tsioutsias (ttsiouts)
11:29:54 efried masber You'll probably want to check your compute logs. Is the compute service started?
11:40:26 efried bauzas naichuans https://review.openstack.org/#/c/514942/ ++! I was poring over this code last night and it made me realize how important this is going to be.
12:01:13 openstackgerrit Surya Seetharaman proposed openstack/nova master: nova-manage cell_v2 verify_instance returns a valid instance mapping even after the instance is deleted/archived https://review.openstack.org/515034
12:13:14 openstackgerrit Eric Fried proposed openstack/nova master: placement: add nested resource providers https://review.openstack.org/377138
12:13:15 openstackgerrit Eric Fried proposed openstack/nova master: placement: allow filter providers in tree https://review.openstack.org/377215
12:13:15 openstackgerrit Eric Fried proposed openstack/nova master: placement: adds REST API for nested providers https://review.openstack.org/384807
12:13:16 openstackgerrit Eric Fried proposed openstack/nova master: placement: update client to set parent provider https://review.openstack.org/385693
12:13:58 efried cdent dansmith Took the liberty of fixing that nit since Jay's still out and the patch is now the first one in the series ^
12:14:19 cdent ✔
12:20:14 cdent efried: on the vgpu stuff, I get that you want to be able to change stuff you can’t express elsewhere, and that we want that
12:20:38 cdent but doesn’t it also make it so the flavor can express VCPU: 2 and then the extra spec can say VCPU: 4
12:21:00 efried cdent That was already the case, intentionally. The extra spec overrides.
12:21:23 cdent I thought the previous behavior was only that the extra spec could set to 0?
12:21:27 efried cdent This was done so that we could move toward the extra spec resources being the single source of truth for all resource requests.
12:21:36 efried uh, stand by...
12:21:50 cdent (for _standard_ classes)
12:22:18 efried cdent No, you could override too.
12:22:41 cdent Basically what I’m trying to understand is if we are making it so there are now two places to do things, and if so, is the plan to end up on only one. If there is, cool.
12:22:42 efried Setting to 0 would remove it. Presumably this is so you could e.g. provide disk resource from somewhere else, but you can't create a flavor with disk 0.
12:23:52 cdent I hope some day we will just destroy flavors as an internal concept and they will only be a UI thing.
12:23:53 efried cdent The plan is to be able to express all resources from one place. We didn't want to (couldn't) do that by adding every standard and custom resource class to the main flavor object. So we want to do it via specific namespacing in the extra_specs because it's a freeform grab-bag.
12:24:44 efried cdent I don't know what the ultimate fate of the flavor object will be. But I don't disagree that it's kind of a mess atm.
12:24:57 efried not least because we're transitioning to this generic resource management ideal.
12:27:18 cdent you got the golden ticket
12:27:25 openstackgerrit Chris Dent proposed openstack/nova master: [placement] Enable limiting GET /allocation_candidates https://review.openstack.org/513526
12:32:31 efried alex_xu yt?
12:32:48 jmccarthy Is there some config or option needed for console.log to work in horizon ? At the moment I'm finding with cirros it works, but with OL7 it doesn't ? The xml of the two instances looks very similar
12:33:08 jmccarthy But in the OL7 case, the console.log is 0 size file
12:33:16 jmccarthy (similar permissions etc)
12:41:05 cdent efried: I hate resorting to mocks in functional tests. (Or really in any tests)
12:41:26 efried cdent Oh, was that a functional test, yeah, don't mock stuff there.
12:41:30 cdent but if you like I’m happy to add that in if we want to be sure, but *shrug*
12:41:39 efried Nah, I +1ed.
12:41:54 cdent k3wl
12:45:18 alex_xu efried: yea, i'm here
12:45:40 alex_xu nova api meeting is in 15 mins at #openstack-meeting-4
12:47:20 efried alex_xu I wanted to talk through that SQL boggle. I'm writing up a comment...
12:49:06 alex_xu efried: yea
12:59:49 sean-k-mooney stephenfin: qq just set up one of my lab vms with an irc client. does my nick ectra look normal. i might set up a znc bouncer at the weekend instead but it windows reboots are getting to be a pain.
13:00:14 stephenfin sean-k-mooney: All good here, yup
13:00:26 sean-k-mooney cool
13:06:16 mriedem dansmith: tonyb gave his blessing https://review.openstack.org/#/c/514685/
13:12:48 openstackgerrit Matt Riedemann proposed openstack/nova stable/pike: Import the config drive docs from openstack-manuals https://review.openstack.org/515053
13:12:48 openstackgerrit Matt Riedemann proposed openstack/nova stable/pike: Import user-data page from openstack-manuals https://review.openstack.org/515054
13:32:37 efried dansmith mriedem This migration UUID business - is that going to get rid of the "doubling up" that makes remove_provider_from_instance_allocation necessary?
13:33:19 mriedem efried: sort of
13:33:41 mriedem efried: the scheduler will pick a dest host and double the instance allocations between the source and dest host,
13:33:48 mriedem then conductor will swap the source node allocation with the migration record
13:34:02 mriedem so the instance will have an allocation on the dest host, and the migration will have an allocation on the source ohst
13:34:04 mriedem *host
13:34:46 mriedem as i pointed out in https://review.openstack.org/#/c/506420/ this morning, until we have cdent's POST allocations stuff, we'll actually be quadrupling the allocations temporarily when resizing to the same host
13:34:54 dtantsur hi mriedem! do I get it right that we're past spec freeze? people ping me re https://review.openstack.org/#/c/449155/, do they have to move it to Rocky at this point? I also wonder if that even needs a spec, but dunno
13:35:52 mriedem dtantsur: i think any move operation involving the ironic driver is going to require a spec yes, simply because of the resource tracking considerations that have to be made with the 1:M host:node mapping with ironic
13:36:03 efried mriedem Okay, so there will still need to be some affordance for adjusting doubled (or quadrupled) allocations for the forseeable future.
13:36:15 mriedem dtantsur: because we have assumptions all over nova when getting compute nodes from the db with only the host that there is only 1 node because we're not supporting moves with ironic yet
13:36:28 mriedem efried: yes, per the FIXME note from dan in that change
13:36:38 mriedem the FIXME means replace that with cdent's POST allocations stuff
13:36:54 cdent just in the midst of writing the docs for that stuff
13:36:58 efried mriedem The problem is in the way we're discovering how much adjustment to make, and to which RP. At the moment the code assumes in several ways that there's only one RP - the compute node.

Earlier   Later