| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2018-10-26 | |||
| 16:06:56 | openstackgerrit | Martin Midolesov proposed openstack/nova master: vmware:PropertyCollector for caching instance properties https://review.openstack.org/608278 | |
| 16:14:36 | openstackgerrit | Martin Midolesov proposed openstack/nova master: VMware: Expose esx hosts to Openstack https://review.openstack.org/613626 | |
| 16:15:11 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Migrate "reboot an instance" user guide docs https://review.openstack.org/612730 | |
| 16:17:28 | mriedem | fried_rice: gibi: do you see something here that i don't? https://review.openstack.org/#/c/605785/16/nova/tests/functional/test_servers.py@5275 | |
| 16:17:49 | mriedem | if i'm wrong, i'll +W | |
| 16:18:23 | openstackgerrit | Merged openstack/python-novaclient master: Fixes Python3 issue in decoding password https://review.openstack.org/604870 | |
| 16:19:55 | fried_rice | mriedem: I didn't scrutinize tbh. If it is a redundant test, I figgered it could be removed in a fup. | |
| 16:21:54 | mriedem | btw, on that safe_connect thing, couldn't all of the same arguments be made for all other services nova talks to if we're worried about network drops? so that would be at least: keystone, glance, cinder, neutron, *barbican | |
| 16:22:06 | mriedem | i think the ironic driver already has it's own retry wrapper | |
| 16:22:24 | cdent | yes? | |
| 16:23:09 | cdent | we assume a lot that's can to be less easy to assume in this like edge and other blech like that | |
| 16:23:15 | cdent | s/can/going/ | |
| 16:23:32 | mriedem | oh i'm sure | |
| 16:23:54 | mriedem | ksa already has retry stuff in it too right? | |
| 16:23:57 | mordred | yup | |
| 16:24:01 | melwitt | ||
| 16:24:16 | mordred | and in sdk we have a structure for expressing per-service error codes that are safe to do automatic retries on | |
| 16:24:39 | cdent | this whole day has been a slow setup to allow mordred to do an sdk advertisement | |
| 16:24:46 | mordred | cdent: isn't htat what most days are? | |
| 16:24:51 | mriedem | we just need to rip and replace all nova's internal client things with the sdk, done! | |
| 16:25:09 | mordred | coming soon to a gerrit near you! | |
| 16:26:03 | mordred | (honestly, I've got one more patch i need to write sdk-side, then I want to make y'all a couple of sake-of-argument patches so we can look at what it might be like and whether it's worth working on in earnest) | |
| 16:27:12 | mriedem | working on what? sdk in nova? | |
| 16:27:31 | mriedem | back in the first YVR summit i think dean said the service projects shouldn't use the sdk, but that was a long time ago | |
| 16:27:50 | mriedem | i think for perf reasons | |
| 16:27:56 | mriedem | but it seems kinda dumb not to | |
| 16:29:06 | leakypipes | gibi, mriedem: anyone got the link to gibi's patch series that removes the doubled-up allocation cruft for legacy evacuate code paths? | |
| 16:29:14 | leakypipes | can't find it atm | |
| 16:30:12 | mriedem | https://review.openstack.org/#/c/512623/ ? | |
| 16:30:42 | mriedem | otherwise idk what that is | |
| 16:30:51 | openstackgerrit | Merged openstack/nova master: Bump os-brick version to 2.6.1 https://review.openstack.org/611109 | |
| 16:31:03 | openstackgerrit | Merged openstack/nova master: Add API ref guideline for examples https://review.openstack.org/604060 | |
| 16:31:16 | melwitt | mriedem: s-1 release review is here. sorry I forgot to do it yesterday https://review.openstack.org/613636 | |
| 16:34:10 | mriedem | done | |
| 16:34:50 | melwitt | thx | |
| 16:41:37 | melwitt | mnaser: do you ever use beta releases of nova? as of http://lists.openstack.org/pipermail/openstack-dev/2018-September/135088.html beta releases are optional and I wasn't sure if operators use them or not | |
| 16:42:56 | melwitt | I proposed a s-1 release out of habit and release folk asked if anyone needed it and to consider not doing a beta release. so just trying to find out if operators we know use them | |
| 16:49:23 | mnaser | I haven’t ever really played with them. Only RCs | |
| 16:49:50 | melwitt | ok, thanks | |
| 17:01:06 | openstackgerrit | Stephen Finucane proposed openstack/nova master: tox: Stop build *all* docs in 'docs' https://review.openstack.org/613645 | |
| 17:01:28 | finucannot | fried_rolls: You'd probably appreciate that ^ | |
| 17:02:41 | openstackgerrit | Stephen Finucane proposed openstack/nova master: Fail to live migration if instance has a NUMA topology https://review.openstack.org/611088 | |
| 17:28:45 | bexigastack | hello all | |
| 17:28:59 | bexigastack | i have a quick question that someone here could possibly answer quickly | |
| 17:29:35 | bexigastack | i've got cinder handling nfs mounts served from a netapp in c-mode | |
| 17:29:45 | bexigastack | everything is working as expected | |
| 17:30:15 | bexigastack | i am having an issue when launching an instance on a compute node where that nfs share is already mounted | |
| 17:30:39 | bexigastack | nova attempts to mount it before launching a second instance but is hit with: | |
| 17:31:13 | bexigastack | Command: mount -t nfs nfsserver.dns.name:/OPENSTACK_MOUNT/OPENSTACK_CLOUD_1 /var/lib/nova/mnt/787b0fb4b865db5b3b3c9d305128af32 | |
| 17:31:31 | bexigastack | Exit code: 32 | |
| 17:31:35 | bexigastack | Stdout: u'' | |
| 17:31:39 | bexigastack | Stderr: u'mount.nfs: /var/lib/nova/mnt/787b0fb4b865db5b3b3c9d305128af32 is busy or already mounted\n' | |
| 17:32:11 | bexigastack | oh woops - i think i am in the wrong channel - this is used for development | |
| 17:32:18 | bexigastack | appologies - i will use #openstack | |
| 18:31:30 | openstackgerrit | Merged openstack/nova master: Reject forced move with nested source allocation https://review.openstack.org/605785 | |
| 18:31:38 | openstackgerrit | Merged openstack/nova master: api-ref: Remove unnecessary minimum microversion https://review.openstack.org/613226 | |
| 18:31:46 | openstackgerrit | Merged openstack/nova master: Add a hacking rule for deprecated assertion methods https://review.openstack.org/613184 | |
| 18:52:43 | melwitt | SteelyDan, leakypipes: are either of you willing to please review this nova-status upgrade check that warns if the deployment is using consoles to set the [workarounds]enable_consoleauth = True config option if performing a live/rolling upgrade? https://review.openstack.org/611214 we need to backport it to stable/rocky and were hoping to include it in stable/rocky release as soon as we can | |
| 18:54:39 | melwitt | mnaser: would you like to have a look at the reno update for the nova-consoleauth thing? https://review.openstack.org/613174 | |
| 18:55:25 | melwitt | and let me know if you think it's good | |
| 18:57:20 | leakypipes | melwitt: you see the py3k test failures on that? | |
| 19:01:06 | openstackgerrit | Dan Smith proposed openstack/nova master: Minimal construct plumbing for nova list when a cell is down https://review.openstack.org/567785 | |
| 19:01:06 | openstackgerrit | Dan Smith proposed openstack/nova master: Modify get_by_cell_and_project() to get_not_deleted_by_cell_and_projects() https://review.openstack.org/607663 | |
| 19:01:07 | openstackgerrit | Dan Smith proposed openstack/nova master: Minimal construct plumbing for nova show when a cell is down https://review.openstack.org/591658 | |
| 19:01:07 | openstackgerrit | Dan Smith proposed openstack/nova master: Refactor scatter-gather utility to return exception objects https://review.openstack.org/607934 | |
| 19:01:08 | openstackgerrit | Dan Smith proposed openstack/nova master: Minimal construct plumbing for nova service-list when a cell is down https://review.openstack.org/584829 | |
| 19:01:13 | melwitt | leakypipes: yes, they are unrelated, ssh timeout | |
| 19:01:22 | leakypipes | melwitt: sure? | |
| 19:01:48 | melwitt | leakypipes: yes, why? | |
| 19:02:05 | leakypipes | melwitt: ok, lemme take another look | |
| 19:02:37 | melwitt | the code change is in the 'nova-status upgrade check' CLI command which isn't running in either of those tempest tests | |
| 19:18:03 | openstackgerrit | Jack Ding proposed openstack/nova master: Add I/O Semaphore to limit concurrent disk ops https://review.openstack.org/609180 | |
| 19:32:37 | openstackgerrit | Merged openstack/nova master: Add more documentation for online_data_migrations CLI https://review.openstack.org/605836 | |
| 19:35:33 | cfriesen | mriedem: sean-k-mooney: gibi: question about the HPET spec. given that the HPET removal was not for any security reasons, and that HPET is supported by the minimum versions of qemu/libvirt, can we simplify the feature and just specify something like "hw:hpet" in the flavor/image with no way to disable it (since there's no useful reason to prevent it) | |
| 19:35:58 | cfriesen | no way for the operator to disable/block it, I mean | |
| 19:54:49 | openstackgerrit | Dan Smith proposed openstack/nova master: Modify get_by_cell_and_project() to get_not_deleted_by_cell_and_projects() https://review.openstack.org/607663 | |
| 19:54:50 | openstackgerrit | Dan Smith proposed openstack/nova master: Refactor scatter-gather utility to return exception objects https://review.openstack.org/607934 | |
| 19:54:50 | openstackgerrit | Dan Smith proposed openstack/nova master: Minimal construct plumbing for nova list when a cell is down https://review.openstack.org/567785 | |
| 19:54:51 | openstackgerrit | Dan Smith proposed openstack/nova master: Minimal construct plumbing for nova service-list when a cell is down https://review.openstack.org/584829 | |
| 19:54:51 | openstackgerrit | Dan Smith proposed openstack/nova master: Minimal construct plumbing for nova show when a cell is down https://review.openstack.org/591658 | |
| 21:21:16 | openstackgerrit | Matt Riedemann proposed openstack/nova-specs master: Expose virtual device tags in REST API https://review.openstack.org/393930 | |
| 21:29:51 | mriedem | cfriesen: idk, i'd have to context switch the hpet stuff all back into my brain | |
| 21:29:57 | openstackgerrit | Matt Riedemann proposed openstack/nova-specs master: Expose virtual device tags in REST API https://review.openstack.org/393930 | |
| 21:31:35 | mriedem | cfriesen: if it's as simple as just adding a new image property b/c it's not some kind of security related issue people would want to disable, and maybe (or maybe not) tied to arch of the guest image, then it seems just having an image property would be OK, expecially since it sounds like it's really just a use case for legacy OSes | |
| 21:31:46 | mriedem | i'm not sure why it would even need to be a flavor extra spec really | |
| 21:34:31 | mriedem | that would probably be better than doing a bunch of discovery to determine if hpet is possible on a host and exposing that as yet another trait | |
| 21:37:22 | cfriesen | at least for qemu it doesn't seem to have any security implications as there are other more accurate time sources anyway. I think an image property makes the most sense, and I don't think any discovery is needed | |
| 21:37:53 | mriedem | os_wind_river_special_knob_hpet=true seems ok to me | |
| 21:37:57 | cfriesen | lol | |
| 21:38:25 | cfriesen | okay. I'll talk to Jack, we'll respin with it as simple as possible and the justification for the simplicity | |
| 21:40:07 | cfriesen | one thing...can we specify the hypervisor type somehow? because we do need to land on a qemu node | |
| 21:40:43 | mriedem | isn't that an existing image property? | |
| 21:40:57 | mriedem | https://docs.openstack.org/glance/latest/admin/useful-image-properties.html | |
| 21:41:05 | mriedem | hypervisor_type | |
| 21:41:18 | mriedem | "Note that qemu is used for both QEMU and KVM hypervisor types." | |
| 21:41:20 | mriedem | hmm | |
| 21:41:25 | mriedem | is that a problem? | |
| 21:41:31 | mriedem | i.e. is a kvm node a problem? | |
| 21:42:06 | cfriesen | no, qemu/kvm is fine | |
| 21:42:12 | mriedem | then i think you're covered | |