| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2017-02-07 | |||
| 13:52:38 | johnthetubaguy | pumaranikar: sometimes these errors raised by placement are almost expected, like not found error when you check something is deleted, etc | |
| 13:52:42 | mdbooth | tdurakov: If you'd asked me yesterday, maybe :) | |
| 13:53:35 | pumaranikar | johnthetubaguy, yes. I thought of error level logging since it is in exception block | |
| 13:53:43 | tdurakov | paul-carlton2 eliqiao shaohe_feng pkoniszewski yuntongjin kashyap kmartin claudiub tobasco abhishekk mrhillsman raj_singh wznoinsk siva_krishnan, folks, let's skip the l-m meeting this week. | |
| 13:53:59 | johnthetubaguy | pumaranikar: yeah, need to think about the consumer of the message, it doesn't rally help them in this case | |
| 13:54:20 | johnthetubaguy | pumaranikar: but a debug log might be a good idea, if we include information on the specific exception | |
| 13:55:03 | johnthetubaguy | tdurakov: did you see this: https://review.openstack.org/#/c/429798/ | |
| 13:55:14 | johnthetubaguy | tdurakov: trying to get it into RC2, if possible | |
| 13:56:19 | pumaranikar | johnthetubaguy, ok. i will have look at your comment and make change accordingly | |
| 13:56:55 | tdurakov | johnthetubaguy: I'll review it in 1.5 hours, or earlier | |
| 13:57:54 | johnthetubaguy | tdurakov: I was wondering if the folks in the meeting agree if we should try get that into RC2 I guess | |
| 13:59:32 | tdurakov | johnthetubaguy: btw, have you seen this one https://review.openstack.org/#/c/408002 | |
| 14:00:01 | johnthetubaguy | tdurakov: yeah, I need to look at that one again | |
| 14:00:23 | johnthetubaguy | tdurakov: thats a new option though, not suitable past feature freeze, but good for pike | |
| 14:05:59 | markus_z | Is the placement service mandatory? (I'm still living under a rock :/ ) | |
| 14:07:19 | johnthetubaguy | markus_z: it is for ocata | |
| 14:07:39 | johnthetubaguy | markus_z: n-cpu should fail to start if its not configured with the placement URL now | |
| 14:08:32 | markus_z | johnthetubaguy: Hm, I think that's the issue in our s390 CI. We don't start the placement service (because we have an outdated version of devstack-gate). | |
| 14:10:05 | markus_z | johnthetubaguy: Even if it is not the root-cause, good to know that we are behind (again). I have to fix that. | |
| 14:10:28 | markus_z | johnthetubaguy: dammit, yeah, that's it | |
| 14:10:31 | markus_z | Thanks! | |
| 14:10:56 | johnthetubaguy | np | |
| 14:11:17 | openstackgerrit | Trevor McCasland proposed openstack/nova-specs master: Add spec for Accelerating SR-IOV with DPDK https://review.openstack.org/429867 | |
| 14:13:07 | openstackgerrit | John Garbutt proposed openstack/nova master: Default live_migration_progress_timeout to off https://review.openstack.org/429798 | |
| 14:13:20 | openstackgerrit | John Garbutt proposed openstack/nova master: WIP: Remove stuck live-migration detection https://review.openstack.org/430218 | |
| 14:14:28 | openstackgerrit | Pushkar Umaranikar proposed openstack/nova master: Placement api: set custom json_error_formatter in allocations https://review.openstack.org/395194 | |
| 14:16:46 | johnthetubaguy | pumaranikar: thanks | |
| 14:18:42 | pumaranikar | johnthetubaguy, thanks for quick review | |
| 14:25:13 | openstackgerrit | Ed Leafe proposed openstack/nova master: Add check for invalid inventory amounts https://review.openstack.org/428800 | |
| 14:25:14 | openstackgerrit | Ed Leafe proposed openstack/nova master: Add check for invalid allocation amounts https://review.openstack.org/416752 | |
| 14:27:17 | openstackgerrit | Stephen Finucane proposed openstack/nova master: conf: Deprecate two more nova-net options https://review.openstack.org/412390 | |
| 14:27:17 | openstackgerrit | Stephen Finucane proposed openstack/nova master: trivial: Standardize indentation of test_vif https://review.openstack.org/430271 | |
| 14:27:18 | openstackgerrit | Stephen Finucane proposed openstack/nova master: trivial: Standardize naming of variables https://review.openstack.org/430272 | |
| 14:28:21 | openstackgerrit | Silvan Kaiser proposed openstack/nova master: Removes getfattr from Quobyte Nova driver https://review.openstack.org/428646 | |
| 14:38:29 | johnthetubaguy | sfinucan: doesn't allow_same_net_traffic get used by neutron, or do we use a different firewall driver? | |
| 14:38:45 | sfinucan | johnthetubaguy: What's that in context to? | |
| 14:39:53 | sfinucan | johnthetubaguy: I'm pretty sure it's mainly used by nova-network, but libvirt also overloads it for its own firewall | |
| 14:40:10 | johnthetubaguy | sfinucan: https://review.openstack.org/#/c/412390/6/nova/conf/network.py@957 | |
| 14:40:55 | sfinucan | johnthetubaguy: Yeah, so my reply to the first option also stands for the second one | |
| 14:41:16 | sfinucan | We're overloading the value for libvirt's firewall and that option should be either (a) removed altogether or (b) moved to the libvirt group | |
| 14:41:32 | johnthetubaguy | sfinucan: agreed with the first one, but trying to workout what does the neutron firewalling? isn't it using this code too? https://github.com/openstack/nova/blob/d5f113de34589933c23a7cecc25ca0bf5f47207d/nova/virt/libvirt/firewall.py#L165 | |
| 14:42:01 | johnthetubaguy | sfinucan: I think its shared in virt via this one though: https://github.com/openstack/nova/blob/4f91ed3a547965ed96a22520edcfb783e7936e95/nova/virt/firewall.py#L307 | |
| 14:42:18 | johnthetubaguy | sfinucan: I guess the others maybe don't use it for any neutron drivers yet, unsure | |
| 14:43:24 | sfinucan | johnthetubaguy: Hmm, that's a good point | |
| 14:45:23 | sfinucan | johnthetubaguy: If we're reading that right, the comments for 'use_ipv6' are also wrong https://github.com/openstack/nova/blob/d5f113de34589933c23a7cecc25ca0bf5f47207d/nova/virt/libvirt/firewall.py#L158 | |
| 14:45:49 | sfinucan | https://github.com/openstack/nova/blob/d5f113de34589933c23a7cecc25ca0bf5f47207d/nova/conf/netconf.py#L68-L70 | |
| 14:45:53 | johnthetubaguy | sfinucan: yeah, it looks that way | |
| 14:46:26 | sfinucan | grr, whoever thought it was a good idea to reuse nova-net options for neutron deserves a smack | |
| 14:46:27 | sfinucan | :) | |
| 14:51:57 | sfinucan | And use_ipv6 is *also* used for that injected network template (compute.injected_network_template). Ugh... | |
| 15:00:28 | lyarwood | https://review.openstack.org/#/c/427364/ - I'd appreciate reviews on this swap_volume bugfix if any cores have time today. sfinucan, mdbooth's NFS fix now depends on this btw. | |
| 15:02:47 | sfinucan | lyarwood: I'll take a gawk shortly | |
| 15:02:55 | lyarwood | sfinucan: cheers | |
| 15:05:01 | openstackgerrit | Jordan Pittier proposed openstack/nova master: api-ref: document the 'tenant_id' query parameter https://review.openstack.org/430290 | |
| 15:06:15 | lyarwood | hmmmm there's no #openstack-devstack channel, does anyone know if openstack-infra is the best place to ask for devstack reviews? I'm looking for reviews on https://review.openstack.org/#/c/427602/ so our ceph job can pass again... | |
| 15:06:41 | dansmith | lyarwood: -qa | |
| 15:07:01 | lyarwood | dansmith: thanks | |
| 15:20:53 | sfinucan | johnthetubaguy: Out of curiosity, what's the finger in the air estimate for cells v1 going bye bye? | |
| 15:21:00 | sfinucan | ...and, in turn, nova-network | |
| 15:21:05 | openstackgerrit | Christopher Martin proposed openstack/nova master: Stop using mox in test_compute_cells.py https://review.openstack.org/389750 | |
| 15:22:01 | johnthetubaguy | sfinucan: we need to implement the transition code, then let that cook for a release. | |
| 15:22:12 | johnthetubaguy | sfinucan: after we actually support two cells v2 cells | |
| 15:22:30 | johnthetubaguy | sfinucan: so another two or three cycles till we delete, maybe :( | |
| 15:22:41 | sfinucan | ouch | |
| 15:23:04 | johnthetubaguy | sfinucan: looking after our users is expensive | |
| 15:23:14 | sfinucan | johnthetubaguy: aye, and sadly necessary | |
| 15:32:55 | dansmith | sfinucan: but we should only maintain n-net for cellsv1 users with legit bugs I think, not just anyone | |
| 15:34:10 | sfinucan | dansmith: I was hoping for an easy way out of unravelling the nest of conf options that are shared between nova-net and libvirt/Xen/other things | |
| 15:34:25 | sfinucan | namely, wait until nova-net is gone then remove the ones we don't have references for anymore :) | |
| 15:34:29 | dansmith | heh | |
| 15:37:15 | sfinucan | johnthetubaguy: So I'm figuring out what nova-network options /have been/ repurposed and duplicate them in other groups with UPGRADE renos | |
| 15:37:56 | sfinucan | then I'm going to mass migrate all nova-net opts to a single conf file which we can leave to d̶i̶e̶ mature | |
| 15:38:01 | sfinucan | Sound fair? | |
| 15:38:21 | johnthetubaguy | that sounds like a good plan | |
| 15:40:41 | sfinucan | (y) | |
| 15:46:13 | BobBall | johnthetubaguy: Thanks for the review - are you +2 on the revision to https://review.openstack.org/#/c/424428/ or are further changes needed? | |
| 15:52:51 | openstackgerrit | Balazs Gibizer proposed openstack/nova master: handle uninited fields in notification payload https://review.openstack.org/415857 | |
| 16:06:43 | pkholkin | johnthetubaguy: mriedem: jaypipes: baoli | |
| 16:07:15 | pkholkin | oops | |
| 16:08:46 | pkholkin | johnthetubaguy: mriedem: jaypipes: bauzas: dansmith: alex_xu: and all nova folks, I decided to leave OpenStack, thank you for your work in community and your support! maybe we will meet once again, wish you all the best and good luck! :) | |
| 16:09:19 | dansmith | pkholkin: bummer, but good luck on your next adventure! | |
| 16:09:43 | mriedem | pkholkin: thanks for your contributions and good luck in the new position | |
| 16:09:57 | pkholkin | dansmith: sorry) thank you ! | |
| 16:10:38 | pkholkin | mriedem: thanks! it was a great pleasure to work in our community! | |
| 16:15:31 | rpodolyaka | pkholkin: good luck, Pasha! really sad to see you go :( | |
| 16:16:38 | markus_z | pkholkin: too bad, thanks for your time | |
| 16:17:18 | pkholkin | rpodolyaka: me too( thank you very much, Roman ! | |
| 16:17:26 | pkholkin | markus_z: thanks! | |
| 16:19:39 | mdbooth | ARGH! | |
| 16:20:01 | mdbooth | The detach/delete race isn't fixable on the compute host | |
| 16:20:20 | mdbooth | Because api gets the bdms and sends them to the compute host, which then acts on them | |
| 16:20:26 | mdbooth | Even though they might be out of date | |
| 16:20:37 | mdbooth | Why? | |
| 16:21:11 | mriedem | mdbooth: did you find my comment in the compute manager code about that already? | |
| 16:21:34 | mdbooth | mriedem: Not yet. I have spelunked significantly, but not everywhere. | |
| 16:21:46 | mriedem | https://github.com/openstack/nova/blob/46c778efafa3ff3c7bb99d74656bb2d2869ceb16/nova/compute/manager.py#L2413 | |
| 16:22:00 | mdbooth | mriedem: Incidentally, while I'd like to fix this race, I think it's orthogonal to the NFS thing. | |
| 16:22:08 | mdbooth | Because the NFS change is safe either way. | |
| 16:22:22 | mdbooth | It's just ugly complaining about removing things which no longer exist. | |
| 16:23:10 | mdbooth | mriedem: Why not just ignore what api gave us and refresh anyway? | |