| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2018-05-11 | |||
| 13:12:38 | zzzeek | mriedem: i could argue this is a bug, b.c. the problem is, nova generating URLs into the database makes it very inconvenient to add new parameters to the URL | |
| 13:14:42 | mriedem | since this doesn't change default behavior yes we could potentially make an exception here for the cells database connection URL stuff; it would be nice if there were a bug associated with this | |
| 13:14:54 | mriedem | i've -2d the ocata backport in oslo.db because you skipped queens and pike | |
| 13:15:56 | stephenfin | mriedem: Think you could have a look at https://review.openstack.org/#/q/topic:bump_min_libvirt_and_qemu+status:open today? | |
| 13:16:13 | mriedem | stephenfin: i sure can | |
| 13:16:22 | stephenfin | thanking you kindly | |
| 13:18:58 | mriedem | melwitt: i've posted release requests for all stable branches and added you as reviewer | |
| 13:26:03 | openstackgerrit | Vladyslav Drok proposed openstack/nova master: Revert "libvirt: Report the allocated size of preallocated file based disks" https://review.openstack.org/567858 | |
| 13:26:28 | openstackgerrit | Vladyslav Drok proposed openstack/nova master: Revert "libvirt: Report the allocated size of preallocated file based disks" https://review.openstack.org/567858 | |
| 13:27:19 | mriedem | vdrok: gdi https://review.openstack.org/#/q/23bd8f62634707fc9896a38ff4dae606c89c6c4b | |
| 13:28:11 | vdrok | hey mriedem :) | |
| 13:28:19 | mriedem | lyarwood: you probably want to look at ^ | |
| 13:29:42 | mriedem | vdrok: i wonder if it would be possible to show a recreate of this in our live migration job simply by changing config_drive_format=vfat in nova.conf in devstack | |
| 13:30:02 | vdrok | mriedem: yup it should be possible | |
| 13:30:26 | mriedem | probably also need to set CONF.libvirt.images_type=raw | |
| 13:31:40 | mriedem | vdrok: i'll push a change to see | |
| 13:31:49 | mriedem | now that we have the live migration job defined in tree \o/ | |
| 13:32:50 | vdrok | also I guess changing https://github.com/openstack/nova/blob/master/playbooks/legacy/nova-live-migration/run.yaml#L31 to 1 | |
| 13:33:00 | mriedem | yes good point | |
| 13:33:09 | mriedem | i'm not actually sure why that is disabled in the multinode jobs | |
| 13:33:15 | lyarwood | I don't see what that commit has to do with the error tbh | |
| 13:34:18 | lyarwood | vdrok: what is it you think the commit is doing to cause that libvirt error? | |
| 13:34:40 | vdrok | lyarwood: my thought was that libvirt utils create_image uses the allocated size (like 220K) in qemu-img create instead of 64mb | |
| 13:35:59 | vdrok | https://github.com/openstack/nova/blob/master/nova/virt/libvirt/driver.py#L7652 | |
| 13:37:01 | openstackgerrit | Matt Riedemann proposed openstack/nova master: DNM: test raw image live migration with vfat config drive https://review.openstack.org/567860 | |
| 13:37:03 | mriedem | vdrok: ^ if that fails, we can then run it with the revert | |
| 13:37:43 | vdrok | yup thank you mriedem | |
| 13:38:39 | openstackgerrit | Kashyap Chamarthy proposed openstack/nova master: libvirt: Remove usage of migrateToURI2() API https://review.openstack.org/567258 | |
| 13:51:39 | mriedem | stephenfin: should probably also have sean-k-mooney look at https://review.openstack.org/#/c/565471/ | |
| 13:51:49 | mriedem | i think one of the intel 3rd party CIs uses vhostuser too...i'll check | |
| 13:52:43 | stephenfin | Not a bad idea. I'd like sahid to look at it too. Hopefully he's back next week | |
| 13:52:57 | mriedem | hmm | |
| 13:52:57 | mriedem | tempest-dsvm-multinode-ovsdpdk-nfv-networking-xenial tempest-dsvm-multinode-ovsdpdk-nfv-networking-xenial : SKIPPED (non-voting)tempest-dsvm-ovsdpdk-nfv-networking-xenial tempest-dsvm-ovsdpdk-nfv-networking-xenial : SKIPPED (non-voting) | |
| 13:56:25 | mriedem | stephenfin: since you're not passing a plugin kwarg to construct the VIFHostUser object, https://review.openstack.org/#/c/565471/4/nova/network/os_vif_util.py@408 - which plugin gets used by default? | |
| 14:01:04 | lyarwood | vdrok / mriedem ; FWIW I'm against reverting that change, the underlying issue appears to be trivial, I'll propose something now. | |
| 14:01:56 | vdrok | lyarwood: that works for me as well :) thank you | |
| 14:03:37 | openstackgerrit | Chris Dent proposed openstack/nova master: Optional separate database for placement API https://review.openstack.org/362766 | |
| 14:04:08 | openstackgerrit | Chris Dent proposed openstack/nova master: Use nova.db.api directly https://review.openstack.org/543262 | |
| 14:04:22 | openstackgerrit | Chris Dent proposed openstack/nova master: Isolate placement database config https://review.openstack.org/541435 | |
| 14:05:25 | openstackgerrit | Chris Dent proposed openstack/nova master: WIP: Ensure that os-traits sync is attempted only at start of process https://review.openstack.org/553857 | |
| 14:05:45 | mriedem | stephenfin: looks ok except i'm not sure which plugin is used to plug a generic vhostuser interface | |
| 14:06:19 | needsleep | Hey, for https://review.openstack.org/#/c/563722/ do you guys want a reno added? | |
| 14:06:36 | melwitt | mriedem: cool about nova-slow. will monitor the ML thread | |
| 14:07:01 | mriedem | needsleep: that's a question for Guest82609 | |
| 14:12:50 | mriedem | figleaf: cdent: probably going to need some API SIG input here https://review.openstack.org/#/c/567687/1/doc/source/contributor/api.rst@290 | |
| 14:14:02 | cdent | mriedem: 410 seems a good idea to me, but will check the law books before commenting | |
| 14:14:12 | mriedem | ok, and it's what we've already used | |
| 14:23:30 | stephenfin | mriedem: Yeah, I'm trying to figure that out now. I'd assumed it was defaulting to a plugin, which is what it does for a missing profile, but I can't find anything to suggest that. I think we'll need sean-k-mooney alright | |
| 14:23:43 | stephenfin | mriedem: I'll rebase the other two patches there to remove the dependency | |
| 14:26:05 | sean-k-mooney | stephenfin: you called :) | |
| 14:26:18 | stephenfin | sean-k-mooney: oh hai | |
| 14:26:38 | sean-k-mooney | i litrally just got to my desk so have no idea what you are talking about | |
| 14:26:55 | stephenfin | sean-k-mooney: I'm not sure if I've been very clever or very dumb in patch. Could you have a look, if you have time? https://review.openstack.org/#/c/565471/ | |
| 14:27:03 | melwitt | fried_rice: done | |
| 14:27:12 | fried_rice | melwitt: Thank you | |
| 14:27:18 | stephenfin | Also | |
| 14:27:24 | sean-k-mooney | stephenfin: oh generic vhost-user | |
| 14:27:37 | finucannot | Yeah, is that a thing? | |
| 14:27:37 | sean-k-mooney | in os-vif i was ment to create a new plugin for that | |
| 14:28:03 | sean-k-mooney | finucannot: ya generic vhost user is used by vpp and snabb switch | |
| 14:29:08 | sean-k-mooney | finucannot: in both cases they dont need any plug operation because teh backeds watch for the sockets thems selves and conect. | |
| 14:29:41 | sean-k-mooney | finucannot: they get the socket info from there neutron ml2 drivers instead of nova | |
| 14:32:14 | sean-k-mooney | finucannot: do you want me to create the generic vhost-user plugin in os-vif so you can delete the get_config_vhostuser stuff? | |
| 14:33:42 | finucannot | Yup, that sounds like a good idea. I'd been looking at snabb switch alright http://blog.allenx.org/2015/07/13/the-new-feature-vhost-user-in-qemu | |
| 14:35:09 | sean-k-mooney | finucannot: i was talking to superdan about this a week or two ago but i was then on vacation so havent done it yet | |
| 14:36:53 | superdan | needsleep: why would that need a reno? it's just internal plumbing right? | |
| 14:37:18 | needsleep | superdan: Just asking because dtantsur suggested that it might. | |
| 14:37:35 | superdan | needsleep: okay I don't really think so. I'll comment | |
| 14:37:43 | needsleep | I suspect it would be noise, tbh | |
| 14:37:45 | needsleep | superdan: ack | |
| 14:39:52 | openstackgerrit | Stephen Finucane proposed openstack/nova master: libvirt: MIN_LIBVIRT_VHOSTUSER_MQ https://review.openstack.org/565472 | |
| 14:39:53 | openstackgerrit | Stephen Finucane proposed openstack/nova master: libvirt: Drop MIN_LIBVIRT_PF_WITH_NO_VFS_CAP_VERSION https://review.openstack.org/565517 | |
| 14:49:59 | openstackgerrit | Merged openstack/nova stable/queens: Update docs for [keystone_authtoken] changes since Queens https://review.openstack.org/564336 | |
| 14:52:09 | jgwentworth | lyarwood: I've been meaning to ask you, would you consider native qemu luks decryption to be the resolution for this bug? or is it still a bug to keep open? https://bugs.launchpad.net/nova/+bug/1633033 | |
| 14:52:09 | openstack | Launchpad bug 1633033 in OpenStack Compute (nova) "live migration with encrypted volume fails" [Undecided,In progress] - Assigned to Lee Yarwood (lyarwood) | |
| 14:53:12 | lyarwood | jgwentworth: yeah not directly, there was an earlier change in the series that fixed this, I was sure I commented on that bug a few weeks ago sorry! | |
| 14:53:52 | lyarwood | jgwentworth: https://review.openstack.org/#/c/460243/ | |
| 14:54:23 | jgwentworth | lyarwood: a-ha, thanks. so with that, the bug can be marked Fix Released? someone was asking about it last week I think | |
| 14:55:39 | lyarwood | jgwentworth: yeah I think so | |
| 14:56:03 | jgwentworth | lyarwood: cool, thanks. I'll update it as such | |
| 14:59:02 | sean-k-mooney | finucannot: ill have a basic noop plugin up for review in the next hour or so. just need to add docs and unit tests working but have a scrum meeting in 2 minutes | |
| 14:59:21 | sean-k-mooney | finucannot: ill comment back on your patch when its ready | |
| 15:00:08 | finucannot | sean-k-mooney: Sounds good to me :) | |
| 15:06:48 | vdrok | mriedem: hrm, it seems like images_type is being overridden somewhere http://logs.openstack.org/60/567860/1/check/legacy-grenade-dsvm-neutron-multinode-live-migration/73d8b8a/logs/etc/nova/nova.conf.txt.gz | |
| 15:07:47 | mriedem | vdrok: the live migration job runs first w/o ceph and then with ceph | |
| 15:08:08 | mriedem | https://github.com/openstack/nova/blob/master/nova/tests/live_migration/hooks/ceph.sh#L79 | |
| 15:08:25 | mriedem | but the first run before ceph should fail | |
| 15:08:41 | mriedem | https://github.com/openstack/nova/blob/master/nova/tests/live_migration/hooks/run_tests.sh#L27 | |
| 15:14:38 | openstackgerrit | Zack Cornelius proposed openstack/nova master: Implement file backed memory for instances in libvirt https://review.openstack.org/567876 | |
| 15:16:31 | openstackgerrit | Matt Riedemann proposed openstack/nova master: WIP: xenapi: drop deprecated vif_driver config option https://review.openstack.org/567877 | |
| 15:16:34 | mriedem | naichuans: hi, i'll probably need some help with fixing the unit tests in ^ because of the way the fake / stubs are setup to trace xenserver session records | |
| 15:16:46 | mriedem | that entire fake/stub structure is a bit weird if you're not used to it | |
| 15:21:43 | openstackgerrit | Surya Seetharaman proposed openstack/nova master: Metadata-API fails to retrieve avz for instances created before Pike https://review.openstack.org/567878 | |
| 15:22:11 | mriedem | vdrok: although something else is up, because in the run before rbd, it uses 'default' which defaults to qcow2, not raw http://logs.openstack.org/60/567860/1/check/nova-live-migration/90babb9/logs/screen-n-cpu.txt.gz#_May_11_14_04_55_065852 | |
| 15:23:12 | mriedem | i'll just override the config option defaults in code rather than mess with setting the values in the CI run | |
| 15:24:45 | openstackgerrit | Surya Seetharaman proposed openstack/nova master: Metadata-API fails to retrieve avz for instances created before Pike https://review.openstack.org/567878 | |
| 15:26:43 | openstackgerrit | Matt Riedemann proposed openstack/nova master: DNM: test raw image live migration with vfat config drive https://review.openstack.org/567860 | |
| 15:29:27 | cfriesen_ | sean-k-mooney: sriov question for you. according to the docs at https://docs.openstack.org/neutron/pike/admin/config-sriov.html, if we have both PFs and VFs on the host and pass the PF through to the guest, the VF number stored in "sriov_numvfs" is lost. how do we restore that when we detach the PF from the guest and return it to the host? | |
| 15:30:35 | sean-k-mooney | cfriesen_: udev rules is the only way i have found really | |
| 15:31:11 | sean-k-mooney | you could perhaps use a systemd service and a filepath trigger | |