Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2017-06-20
19:09:06 melwitt but the other stuff isn't in the table. yeah
19:09:53 mriedem mgagne: sounds good
19:10:05 mriedem mgagne: i imagine you may be a bit sore
19:10:43 mgagne all fine, just a bit out of time for today ;)
19:11:27 mriedem melwitt: yeah it's this i think https://github.com/openstack/nova/blob/master/nova/api/openstack/compute/floating_ip_dns.py#L225
19:11:58 melwitt ah, yep, that looks like it
19:12:02 mriedem melwitt: so you can't add a floating ip address dns entry to the dns domain if you don't already have the floating ip address,
19:12:02 mriedem so,
19:12:22 mriedem we can probably just delete the floating ip if we go over-quota, rather than call deallocate_floating_ip
19:12:25 mriedem like you're doing
19:12:43 melwitt yeah, so it just looks like the same deal as the fixed ip alloc/dealloc thing, you can't have made other associations before the REST call returns something to the user for the user to call back and make associations
19:13:06 mriedem ok, and if we mess this up, the .002% of people that will likely hit it can yell at us
19:13:18 melwitt so I probably lucked out on that. and yes, that is also true heh
19:13:20 mriedem we'll just tell them dansmith told us it was ok
19:13:30 melwitt :D
19:13:54 melwitt sdague had a good point about calling that option "strict_quotas" instead of "recheck_quota." now that's gonna bug me
19:14:30 melwitt I know I could technically go ahead and deprecate "recheck_quota" and add strict_quotas and have both work for awhile
19:15:34 sdague melwitt: ++
19:15:48 sdague it's not in a stable yet, so now would be the time to do it
19:15:48 mriedem melwitt: +W on the change,
19:15:57 mriedem you're getting eerily close to seeing light at the end of this quota tunnel
19:16:30 melwitt sdague: yeah, I think I'll do it then. that name is so much better. I just couldn't think of anything good
19:17:28 melwitt mriedem: \o/ can't believe it! also, I need to add all of the recheck quota test cases for instances now, that's gonna be a pain
19:18:36 melwitt I meant to have done it already but was busy and sort of avoiding it :(
19:26:07 mriedem that's ok, i'm putting off supporting compute node paging across multiple cells
19:33:24 sdague melwitt: so, given the bug dan was debugging this morning, is this bit right - https://review.openstack.org/#/c/416521/46/nova/compute/api.py@3022
19:33:48 sdague I also kind of wonder if there is going to need to be some standard scatter/gather callers here
19:34:54 dansmith sdague: I just replied to both
19:35:14 melwitt sdague: it is now wrong since dansmith's change to how target_cell works. the change is really old
19:35:36 melwitt so, good catch
19:40:36 openstackgerrit Merged openstack/nova master: placement: separate normalize_resources_qs_param https://review.openstack.org/475599
19:41:10 openstackgerrit Merged openstack/nova master: Use more specific asserts in tests https://review.openstack.org/466609
19:57:25 openstackgerrit Ildiko Vancsa proposed openstack/nova master: WIP: Add Cinder v3 detach to swap_volume https://review.openstack.org/456971
20:31:47 cfriesen has anyone looked at what the qemu logs look like when block live-migrating under heavy localdisk IO? does it show up in the "Migration running for..." log from the live migration monitor?
20:48:16 mriedem melwitt: were you looking for a release note on this? https://review.openstack.org/#/c/393695/
20:49:30 melwitt mriedem: not really, just wanted a sanity check that we can be assured adding a minimum wouldn't break anyone. maybe a release note is the right thing?
20:50:13 melwitt I don't know why anyone would use -1 but what if they did. I didn't check yet what a -1 would do for that interval, maybe it just wouldn't even work
20:50:32 mriedem melwitt: don't know, but i left a reply in there too - -1 has special meanings in things
20:50:34 mriedem like retry forever,
20:50:35 mriedem or disable
20:51:00 melwitt yeah. so maybe the question is, does -1 work presently or not. if not, then the point is moot
20:57:40 openstackgerrit Merged openstack/nova master: Count floating ips to check quota https://review.openstack.org/457862
21:03:31 openstackgerrit Eric Fried proposed openstack/nova master: WIP: hypervisor_hostname must match get_available_nodes https://review.openstack.org/475926
21:09:10 openstackgerrit Rick Bartra proposed openstack/nova master: Add policy granularity to the Flavors API https://review.openstack.org/449288
21:16:50 openstackgerrit Merged openstack/nova master: Remove translation of log messages https://review.openstack.org/467948
21:17:22 openstackgerrit Merged openstack/nova master: [placement] Update placement devref to modern features https://review.openstack.org/469047
21:47:52 openstackgerrit Dan Smith proposed openstack/nova master: Fix and optimize external_events for multiple cells https://review.openstack.org/445142
21:52:26 openstackgerrit Eric Fried proposed openstack/nova master: hypervisor_hostname must match get_available_nodes https://review.openstack.org/475926
22:03:12 mriedem dansmith: we need to talk about https://review.openstack.org/#/c/391941/
22:03:23 mriedem the easy one first https://review.openstack.org/#/c/391941/50/nova/compute/cells_api.py
22:03:36 mriedem i'm thinking we don't need to be passing that tag kwarg through for cells v1
22:04:06 dansmith totes agree
22:04:22 mriedem ok now the hard one https://review.openstack.org/#/c/391941/50/nova/compute/api.py
22:04:28 mriedem the way artom has this written,
22:04:36 dansmith if we're checking service version and failing,
22:04:41 mriedem if you try to attach a volume with a tag to a shelved offloaded instance, he fails with a 400
22:04:51 dansmith we should just always fail in cells_api if you pass a tag
22:05:03 mriedem good point
22:05:09 mriedem mayhap you'd like to leave that comment
22:05:43 dansmith mayhap I already did
22:05:49 mriedem ok, so the shelved offloaded case
22:06:05 mriedem he's got this failing for a shelved offloaded instance since you can't call reserve_block_device_name on the compute
22:06:10 mriedem which is what's checking the driver flag
22:06:20 mriedem here https://review.openstack.org/#/c/391941/50/nova/compute/manager.py
22:07:02 dansmith ugh, we support regular volume attach for offloaded instances?
22:07:05 mriedem i'm trying to decide if we should support attaching a volume with a tag to a shelved offloaded instance. his argument was it's not good to mysteriously fail to apply the tag when unshelving
22:07:13 mriedem eff yes we do
22:07:18 mriedem HPE MFer
22:07:30 dansmith yeah, I understand his point, but.. oye
22:07:54 mriedem so with attach to a shelve offloaded instance, we prep the bdms (attach them) to the guest when unshelving
22:08:01 mriedem basically server create, but you know
22:08:18 mriedem like bfv for an already booted instance - it's the best
22:08:49 mriedem anyway, my thinking was we could support attaching volumes to a shelved offloaded instance in basically the same way as we treat normal create
22:08:51 dansmith right, but it would suck to have destined your instance to certain doom while offloaded,
22:08:58 dansmith and then be unable to ever boot it ever because you have no supporting hypervisors
22:09:14 mriedem it wouldn't be certain doom
22:09:31 mriedem and the same can be said for creating a server with a tagged bdm
22:09:46 mriedem if it ends up on a compute that doesn't support tagged devices, it's a BuildAbortException
22:09:49 mriedem remember me freaking out about this earlier in the week?
22:10:48 mriedem on normal server create we check here https://github.com/openstack/nova/blob/4e6a7c18b1c8340b9fdc93ee02e56b5a29a99648/nova/compute/manager.py#L1930
22:11:05 mriedem if that fails (it's a vmware host), we abort the build
22:11:05 mriedem fin
22:12:01 mriedem i was thinking we could allow attaching a tagged bdm to a shelved offloaded instance, and during unshelve we basically have the same flag check, and if the compute doesn't support it then we fail the unshelve
22:17:22 dansmith mriedem: well, the options are:
22:17:23 dansmith 1. refuse to attach with a tag to offloaded instances
22:17:26 dansmith 2. allow it and fail on unshelve
22:17:47 Guest96259 ...I guess that's it, right?
22:18:06 mriedem Guest96259: basically
22:19:16 Guest96259 I dunno, #1 seems not that terrible to me, given how much stuff doesn't really work with shelve
22:19:53 mriedem Guest96259: failing in unshelve seems like a can of worms given we don't do anything with the instance if spawn fails
22:20:34 Guest96259 yeah
22:20:39 mriedem we'll revert the task_state,
22:20:39 mriedem record a fault
22:20:41 mriedem but we don't clean anything up, i.e. https://bugs.launchpad.net/nova/+bug/1627694
22:20:49 mriedem i guess you don't want the instance in ERROR state b/c then you can't attempt to unshelve it again
22:21:50 openstack Launchpad bug 1627694 in OpenStack Compute (nova) "unshelving an instance doesn't rollback volumes connections on failure" [Low,In progress] - Assigned to Shoham Peller (shoham-peller)
22:24:48 mriedem mordred: is your new api user ML a place i can ask if people are actually using some of the compute APIs? like shelve?
22:37:14 openstackgerrit Merged openstack/nova master: libvirt: Extract method _guest_add_video_device https://review.openstack.org/470783
22:37:48 openstackgerrit Merged openstack/nova master: libvirt: Extract method _guest_add_pci_devices https://review.openstack.org/470784
22:41:27 openstackgerrit Merged openstack/nova master: libvirt: Extract method _guest_add_watchdog_action https://review.openstack.org/470785

Earlier   Later