| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2017-06-13 | |||
| 19:07:35 | mriedem | cfriesen: yikes | |
| 19:08:08 | mriedem | sdague: any idea if that works with sphinx 1.5.1? https://github.com/openstack/requirements/blob/master/global-requirements.txt#L412 | |
| 19:08:33 | mriedem | sdague: also curious for your opinion on embedding the nested flavor in the server show ouput as a json blob http://paste.openstack.org/show/612457/ | |
| 19:08:58 | mriedem | i personally think that's ugly, but we also do it for volume attachments and extra specs in the flavor-show output | |
| 19:09:43 | cfriesen | I'm not wedded to it, happy to do something better | |
| 19:09:58 | cfriesen | just don't know what other options we have | |
| 19:11:46 | mriedem | cfriesen: i'd prefer to see it like this somehow http://paste.openstack.org/show/612458/ | |
| 19:11:51 | mriedem | but not sure how easy that is | |
| 19:13:15 | sdague | mriedem: I don't yet, still puttering through this | |
| 19:13:44 | sdague | mriedem: I agree, I think it should be unrolled | |
| 19:13:54 | cfriesen | mriedem: that'd make functional tests a bit of a mess, trying to correlate that against the output of "flavor show" | |
| 19:14:18 | cfriesen | I guess we could take the "flavor show" output and mangle it into that format and then compare | |
| 19:14:34 | sdague | um, why isn't this just flavor_ ephemeral | |
| 19:14:41 | sdague | flavor_ram | |
| 19:14:44 | sdague | etc | |
| 19:14:51 | mriedem | could be namespaced | |
| 19:14:52 | mriedem | that's an option | |
| 19:15:19 | sdague | if you are putting the output in the tables, it should be table friendly data | |
| 19:15:45 | sdague | or flavor:ram | |
| 19:15:54 | cfriesen | sdague: I was just following the existing pattern where the "flavor" entry had the name and the id on one line | |
| 19:16:06 | sdague | cfriesen: right, but that was a key | |
| 19:16:08 | mriedem | sure, but now it's more than id | |
| 19:16:12 | sdague | mriedem: ++ | |
| 19:16:47 | mriedem | cfriesen: i'd think it would be pretty easy to transform that to flavor_ namespaced keys in the server resp body object before sending it to the table output formatter | |
| 19:17:03 | mriedem | there might even be a callback function available with the output formatter | |
| 19:17:07 | cfriesen | agreed. you want "flavor_" or "flavor:"? | |
| 19:17:23 | mriedem | hmm | |
| 19:17:29 | sdague | I think flavor: | |
| 19:17:33 | mriedem | you're treading on OS-SRV-EXT territory :) | |
| 19:17:38 | sdague | mriedem: sure | |
| 19:17:46 | mriedem | i'm cool with flavor: | |
| 19:17:52 | mriedem | makes it more namespacey | |
| 19:17:54 | sdague | yep | |
| 19:18:05 | cfriesen | okay, will give that a try | |
| 19:22:06 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Pull out code that builds VIF in _build_network_info_model https://review.openstack.org/466449 | |
| 19:22:07 | openstackgerrit | Matt Riedemann proposed openstack/nova master: WIP: Handle network-changed event for a specific port https://review.openstack.org/465787 | |
| 19:41:01 | mriedem | no artom?! | |
| 19:41:19 | mriedem | kashyap: i assume if i see this in the libvirtd logs, it means it's a persistent domain: "Writing domain status to disk" | |
| 19:41:35 | mriedem | well i already know it's a persistent domain, so nm | |
| 20:02:59 | openstackgerrit | Merged openstack/nova master: Remove old service version check for mitaka https://review.openstack.org/472395 | |
| 20:18:05 | openstack | Launchpad bug 1696125 in OpenStack Compute (nova) "Detach interface failed - Unable to detach from guest transient domain (pike)" [High,Confirmed] - Assigned to Matt Riedemann (mriedem) | |
| 20:18:05 | mriedem | sgordon: you might like this https://bugs.launchpad.net/nova/+bug/1696125/comments/16 - we have a bug where detaching an ethernet tap device takes ~5 minutes to be gone from the guest | |
| 20:18:08 | mriedem | using linuxbridge | |
| 20:18:23 | mriedem | nova waits ~1-2 minutes for the device to be gone before giving up | |
| 20:18:47 | mriedem | i've done the libvirtd log digging | |
| 20:19:04 | mriedem | not sure if this is a problem in libvirt/qemu or if it's expected to take this long | |
| 20:19:09 | mriedem | not a problem for ovs | |
| 20:19:37 | sgordon | mmm, and if i am perfectly honest most of the testing on our end focuses on ovs these days | |
| 20:22:15 | mriedem | we could bump this timeout for tap devices, but that's just a workaround, and i believe the tempest tests would actually timeout before that | |
| 20:23:18 | mriedem | tempest times out just at 5 minutes | |
| 20:23:26 | mriedem | so it'd be a craptacular race to the bottom | |
| 20:24:27 | sgordon | mriedem, i would like to think it should not in fact take > 1 minute but i will track down someone who is actually intelligent to look at it | |
| 20:24:32 | sgordon | :) | |
| 20:25:05 | mriedem | sgordon: fwiw the neutron side reverted the change to use vif_type='tap' with linuxbridge https://review.openstack.org/#/c/472365/ | |
| 20:25:12 | mriedem | interested to know why that would make a difference | |
| 20:25:41 | mriedem | the vif_type is now using 'bridge' again | |
| 20:26:39 | mriedem | which goes through os-vif | |
| 20:26:42 | mriedem | tap does not yet | |
| 20:43:28 | openstackgerrit | Matthew Edmonds proposed openstack/nova master: Use CONF.host for powervm nodename https://review.openstack.org/473975 | |
| 20:48:33 | openstackgerrit | Matthew Edmonds proposed openstack/nova master: [WIP] Deactivate powervm compute until ready https://review.openstack.org/471773 | |
| 21:07:19 | mriedem | claudiub|2: i thought hyper-v supported fibrechannel? https://docs.openstack.org/developer/nova/support-matrix.html#storage_block_backend_fibrechannel | |
| 21:08:05 | mriedem | https://github.com/openstack/nova/blob/master/nova/virt/hyperv/volumeops.py#L364 | |
| 21:09:30 | mriedem | efried: questions in the powervm support matrix change https://review.openstack.org/#/c/470999/ | |
| 21:10:03 | efried | esberglu ^^ | |
| 21:10:07 | efried | Thanks mriedem, looking. | |
| 21:12:30 | openstackgerrit | Merged openstack/nova master: remove ugly local import https://review.openstack.org/450787 | |
| 21:12:33 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Indicate Hyper-v supports fibre channel in support matrix https://review.openstack.org/473986 | |
| 21:15:36 | claudiub|2 | mriedem: yep, it does. | |
| 21:15:57 | mriedem | claudiub|2: ok patched above | |
| 21:15:58 | claudiub|2 | i guess we have to update the matrix. | |
| 21:17:33 | mriedem | already did ^ | |
| 21:19:09 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Remove service control from feature support matrix https://review.openstack.org/473989 | |
| 21:20:10 | mriedem | sdague: dansmith: do you happen to remember what "service control" meant? ^ | |
| 21:20:39 | mriedem | maybe it meant QoS? | |
| 21:20:48 | dansmith | mriedem: literally bringing up and down services, I think. not sure if that meant nova-compute itself, or things like turning xen on and off for a given node | |
| 21:21:01 | dansmith | before my time I think, but never widely supported AFAIK | |
| 21:21:21 | mriedem | oh like the libvirt driver auto-disabling nova-compute... | |
| 21:21:36 | dansmith | well, I think it was more that that | |
| 21:21:41 | mriedem | either way, | |
| 21:21:49 | dansmith | like if libvirt supported it, it might be system("service libvirt-bin stop") kind of thing | |
| 21:22:00 | mriedem | if we don't know wtf it means, probably best to not doc it and if we care about documenting specifics we do that separately | |
| 21:22:06 | dansmith | yeah | |
| 21:22:17 | dansmith | here's a good description: | |
| 21:22:21 | dansmith | "not a thing we do" | |
| 21:23:10 | mriedem | heh, reminds me of https://developer.openstack.org/api-ref/compute/?expanded=start-host-detail#start-host | |
| 21:23:19 | mriedem | "This is not implemented by any in-tree compute drivers and therefore will always fail with a 501 NotImplemented error. Needless to say, it is not recommended to use this API and it is deprecated as of the 2.43 microversion." | |
| 21:23:35 | mriedem | "here is our middle finger, have a nice day" | |
| 21:25:16 | dansmith | heh | |
| 21:25:36 | dansmith | maybe: | |
| 21:32:33 | mriedem | efried: OS-DCF:diskConfig is the auto disk configstuff | |
| 21:32:43 | mriedem | old xen thing that predated cloud-init i think | |
| 21:32:52 | mriedem | johnthetubaguy could probably explain it | |
| 21:33:14 | mriedem | i'd love to remove it from our api code since it's messy | |
| 21:35:28 | efried | hyperv seems to think it has it also. | |
| 21:36:10 | mriedem | well, i don't think it means anything to other drivers | |
| 21:36:25 | mriedem | tempest tests against that request parameter in our gate, with libvirt | |
| 21:36:32 | mriedem | but i don't think it does anything | |
| 21:37:36 | mriedem | yeah grep auto_disk_config | |
| 21:37:43 | efried | Looks like sdague might've been on the road to removing it. | |
| 21:37:43 | mriedem | it's an image property only used by the xen driver | |