Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2017-09-14
19:28:38 openstackgerrit Matt Riedemann proposed openstack/nova stable/pike: Target context when setting instance to ERROR when over quota https://review.openstack.org/504178
19:28:44 bauzas mriedem: except a very old thread where I was already saying all the good I think about the cinder/nova AZ relationship
19:29:04 bauzas mriedem: so I'm just going to release a new patch for that honestly
19:29:13 bauzas patch == revision I mean
19:29:59 mriedem bauzas: http://lists.openstack.org/pipermail/openstack-operators/2017-June/013676.html
19:30:19 bauzas mriedem: okay, I'll reply then, thanks
19:40:18 mriedem sean-k-mooney: nvm found them https://review.openstack.org/#/q/topic:bp/vif-port-profile,n,z
20:02:02 tasker when doing a live migration in a nova newton, in the log on the destination compute host, I get "TypeError: 'NoneType' object does not support item assignment" from nova/network/neutronv2/api.py#L289 -> port_profile[MIGRATING_ATTR] = host
20:04:38 tasker has anyone seen this or have a clue what the function is trying to do? what is port_profile? where is it set and why would it be None?
20:05:40 openstackgerrit Michael Still proposed openstack/nova master: Move nbd commands to privsep. https://review.openstack.org/500351
20:05:41 openstackgerrit Michael Still proposed openstack/nova master: Move xend existence probes to privsep. https://review.openstack.org/495538
20:05:41 openstackgerrit Michael Still proposed openstack/nova master: Move lvm handling to privsep. https://review.openstack.org/495516
20:05:42 openstackgerrit Michael Still proposed openstack/nova master: Cleanup mount / umount and associated rmdir calls https://review.openstack.org/494423
20:05:42 openstackgerrit Michael Still proposed openstack/nova master: Move shred to privsep. https://review.openstack.org/495537
20:05:43 openstackgerrit Michael Still proposed openstack/nova master: Don't shell out to mkdir, use ensure_tree() https://review.openstack.org/492326
20:05:43 openstackgerrit Michael Still proposed openstack/nova master: WIP / Aspirational: we don't need rootwrap any more. https://review.openstack.org/495542
20:05:44 openstackgerrit Michael Still proposed openstack/nova master: Move loopback setup and removal to privsep. https://review.openstack.org/495664
20:05:45 openstackgerrit Michael Still proposed openstack/nova master: Move ploop commands to privsep. https://review.openstack.org/492325
20:05:47 openstackgerrit Michael Still proposed openstack/nova master: Move kpartx calls to privsep. https://review.openstack.org/500354
20:05:47 openstackgerrit Michael Still proposed openstack/nova master: Move the idmapshift binary into privsep. https://review.openstack.org/495541
20:05:48 openstackgerrit Michael Still proposed openstack/nova master: Move blkid calls to privsep. https://review.openstack.org/500398
20:05:48 openstackgerrit Michael Still proposed openstack/nova master: Move libvirts dmcrypt support to privsep. https://review.openstack.org/490737
20:05:49 openstackgerrit Michael Still proposed openstack/nova master: Move the dac_admin privsep code to a new location. https://review.openstack.org/504194
20:05:49 openstackgerrit Michael Still proposed openstack/nova master: Use symbolic names for capabilities, expand sys_admin context. https://review.openstack.org/504193
20:05:50 openstackgerrit Michael Still proposed openstack/nova master: Squash dac_admin privsep context. https://review.openstack.org/504195
20:11:50 mriedem johnthetubaguy: should we revive this nova spec for the neutron port binding API changes for live migration? https://review.openstack.org/#/c/375580/
20:12:54 cdent mriedem: when are we starting back up?
20:13:17 cdent and what’s next/
20:13:34 mriedem neutron was supposed to be 1-3
20:13:43 mriedem 3 was the generic device mgmt stuff
20:13:51 cdent woot 45 minute break!?
20:13:57 mriedem cburgess wants to talk about something
20:14:03 cdent no change then
20:14:05 cburgess Tomorrow is fine.
20:14:09 cburgess Ouch....
20:14:16 cdent cburgess said with love
20:14:18 cburgess I'll remember that cdent
20:14:21 cburgess Yeah yeah.
20:14:21 mriedem we could get started on our friday pile of stuff
20:15:18 mriedem how about we just get started back up at 2:30
20:15:32 cdent good plan
20:37:15 tasker mriedem: I'm inferring that spec might be related to the problems I'm seeing. is that right?
20:39:19 openstackgerrit Matt Riedemann proposed openstack/nova-specs master: Enable cold migration with target host https://review.openstack.org/489031
20:39:32 mriedem tasker: what spec and what problem?
20:40:03 mriedem tasker: oh, um, no probably different issue
20:40:26 mriedem john's spec is for new functionality
20:40:38 tasker ah. the two were coincidentally close.
20:57:38 openstackgerrit Michael Still proposed openstack/nova master: Have one list of reboot task_states https://review.openstack.org/219981
20:58:07 tasker in the line preceding the TypeError, port_profile = p.get(BINDING_PROFILE, {}), this returns None, which implies that p[BINDING_PROFILE] is None itself.
21:00:04 tasker and, occassioanlly, it doesn't have that value, and returns the empty dictionary {} defined in the .get()
21:01:54 tasker it returns {} when there is no "binding:profile" key in the port dictionary
21:02:11 tasker it returns None when there is a "binding:profile", but when it's set to None.
21:02:37 tasker time to find out where the port is defined. hopefully I can find out why it's sometimes not there and other time is None.
21:07:23 tasker well, it's because the port itself has no "profile".
21:07:33 tasker so it looks like my issue is with neutron, not nova.
21:13:37 openstackgerrit Merged openstack/nova-specs master: Enable cold migration with target host https://review.openstack.org/489031
21:17:10 openstackgerrit Lee Yarwood proposed openstack/nova master: libvirt: Refactor encryptor attach and detach calls https://review.openstack.org/460243
21:18:07 tasker was "binding:profile" introduced in newton?
21:18:27 tasker I'm looking at a mitaka cluster and I don't see that on any of my ports. did nova start looking for in newton?
21:18:36 tasker is "binding:profile" a nova thing?
21:25:17 mriedem tasker: that was introduced here https://review.openstack.org/#/c/275073/
21:25:18 mriedem in newton
21:25:25 openstackgerrit Merged openstack/nova master: Revert "Enable test_iscsi_volume in live migration job" https://review.openstack.org/504143
21:27:05 mriedem i don't see why you'd get a NoneType here https://review.openstack.org/#/c/275073/45/nova/network/neutronv2/api.py@289
21:27:41 mriedem since the port_profile should default to {}, UNLESS the port has binding:profile=None...
21:27:46 tasker because of my ports. they are actually define with "binding:profile | None"
21:27:46 mriedem that's probably what's broken
21:27:49 tasker yes
21:27:51 tasker and!
21:27:56 mriedem it should be: port_profile = p.get(BINDING_PROFILE, {}) or {}
21:28:15 mriedem tasker: ok, can you report a bug?
21:28:22 mriedem we'll have to start with the fix on master and get it backported
21:28:25 tasker I discoverd that the ports don't always have binding:profile -- it's presence changes between lookups.
21:28:29 tasker sure.
21:28:42 openstackgerrit Merged openstack/nova master: Default the service version in the notification tests https://review.openstack.org/499957
21:29:38 openstackgerrit Lee Yarwood proposed openstack/nova stable/pike: Handle keypair not found from metadata server using cells https://review.openstack.org/500953
21:36:34 tasker https://bugs.launchpad.net/nova/+bug/1717365
21:36:35 openstack Launchpad bug 1717365 in OpenStack Compute (nova) "binding:profile is None breaks migration" [Undecided,New]
21:36:44 tasker I hope that's descriptive enough.
21:37:02 tasker I also took the liberty of attaching a patch.
21:38:03 mriedem tasker: works for me, thanks
21:38:34 tasker ok .. now why do my ports sometimes have "binding:profile" and sometimes not. this I think is a question for neutron.
21:39:22 mriedem yup
21:40:33 tasker mriedem: thanks for your help with the nova piece
21:40:36 mriedem tasker: oh it might have something to do with pci devices on the instance
21:40:43 tasker eh?
21:41:09 tasker are they magically disappearing and reappearing between each port lookup?
21:41:38 mriedem tasker: was looking at https://review.openstack.org/#/c/275073/45/nova/network/neutronv2/api.py@1037
21:43:12 tasker oh, interesting.
21:43:49 tasker but I'm asking neutron about the port via `port-show` and sometimes it's there and sometimes it's not. even just a few seconds later.
21:44:10 tasker does neutron ask nova about some info when shows a port?
21:44:14 mriedem hmm, not sure, maybe some kind of race
21:44:18 mriedem no
21:44:22 tasker didn't think so.
21:44:56 mriedem i don't know enough about neutron, but could be something overwriting the profile from an agent or something coming in with a late update? idk
21:45:40 tasker yeah. thanks anyway for the point to the pci device.
21:49:02 dansmith mriedem: sdague: am I remembering that there is something non-intuitive about name-based referencing of instances, like that it's really a pattern match and not an exact string match?
21:49:38 mriedem instance_name_template
21:49:39 mriedem ?
21:49:45 mriedem or the actual db api query?

Earlier   Later