| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2020-05-05 | |||
| 20:31:11 | dansmith | I haven't been paying attention here | |
| 20:31:24 | dansmith | but anything that generates db or mq traffic should be authenticated, I think | |
| 20:31:30 | dansmith | even if read-only | |
| 20:31:36 | artom | dansmith, it doesn't | |
| 20:32:03 | dansmith | okay, so a only a healthcheck of "can you respond to http requests" then? | |
| 20:32:12 | artom | dansmith, yep | |
| 20:32:18 | dansmith | because generally those expand to "can you talk to your database?" etc | |
| 20:32:49 | artom | dansmith, yeah, gmann wanted to do that, but IMO it's complexity we don't necessarily need (right now) | |
| 20:33:02 | artom | For instance, which database? api? cell1? What about multicell? | |
| 20:33:36 | dansmith | artom: right, so if it's going to grow to that (which most people will want it it, AFAIK) you should consider it needing to be auth'd | |
| 20:34:04 | gmann | dansmith: yeah, my idea is to expand that to DB checks also | |
| 20:34:06 | artom | dansmith, no argument there from me | |
| 20:34:07 | dansmith | because I would expect returning a dict of each db (api, cell1, cell2, etc) as well as each mq would be exactly what people want | |
| 20:34:16 | dansmith | gmann: right | |
| 20:34:38 | dansmith | artom: so are you arguing that if you have no auth, you get a "yes I'm here but that's all I will tell you" response/ | |
| 20:36:04 | gmann | artom: and i will say we finalize it now what all things we want to do with healthcheck instead of expending it later which make lot of things change like auth, enable/disable things etc. but only https request checks we are like 'everything is fine as it is' | |
| 20:36:10 | artom | dansmith, it was more of a response to gmann's security risk argument - that just knowing a service is running somewhere is not a security risk, or if it is, we've done something horribly wrong | |
| 20:36:48 | artom | gmann, you mean future proofing it? | |
| 20:37:08 | bnemec | We've been discussing them for years and nobody has actually implemented anything. | |
| 20:37:16 | gmann | that is why i suggested to have spec and then discuss what all things we want to do with that. | |
| 20:37:49 | dansmith | gmann: we should never expect to design something like that once and forever, but agree we should not design it before making it useful, only to have to revise | |
| 20:37:55 | artom | gmann, yeah, I get that... It's just... looks to me like it'll get bogged down in the kind of discussion that we're having :) | |
| 20:38:27 | gmann | bnemec: yeah, it has more attention now as zigo also brought it up for doing it for all projects as community goal which is good idea. | |
| 20:38:42 | artom | So I'd be more in the campg of "do something easy and useful *now*, trying to future-proof it", instead "do something perfect later" | |
| 20:38:44 | bnemec | It's not the first time that has happened either. I proposed it in Berlin as a community goal. | |
| 20:38:46 | dansmith | artom: so you leave all your computer's ports open while connected directly to the internet and depend on windows/samba auth to keep people out of your files right? :) | |
| 20:39:23 | dansmith | expose ssh to everyone, relying on your password and the patched-ness of your sshd, etc | |
| 20:40:12 | artom | I mean, I've almost done the latter | |
| 20:40:34 | gmann | artom: but it can break things later say, it is enable by default and if we extend in future for things need more care then we may want to disable it by default. | |
| 20:40:54 | gmann | bnemec: ah, did not know that :) | |
| 20:40:59 | artom | gmann, I'm not arguing off by default :) | |
| 20:41:01 | dansmith | artom: but you wouldn't tell everyone to do that as matter of policy right? there're legit reasons to limit scope and reach, not all of which are "have done something terribly wrong" | |
| 20:42:09 | artom | dansmith, yes, I understand reducing target aread | |
| 20:42:11 | artom | *area | |
| 20:46:15 | melwitt | gmann: just a note that we recently made version discovery unauth'ed https://review.opendev.org/685181 if that is relevant at all | |
| 20:50:14 | gmann | melwitt: yeah that is fine but in case of healthcheck which can be extended on more deeper check of DB etc so keeping it unauth and enable by default with no way to disable is my concern. at least we can make it disable by default and let operator to enable it knowingly. | |
| 20:50:27 | melwitt | k | |
| 20:50:30 | gmann | with good help msg in config option. | |
| 20:50:57 | dansmith | agree | |
| 21:08:36 | zigo | artom: I think writing what you describe as the perfect /healthcheck should prevent enabling what we already have. | |
| 21:09:11 | artom | zigo, wait, why is that addressed to me? I'm agreeing with bnemec and you, somethign good now is better than something perfect later :) | |
| 21:09:19 | zigo | Don't get me wrong, I'd love to have what you described. | |
| 21:09:22 | zigo | Oh, wrong person? :) | |
| 21:09:31 | artom | dansmith I think | |
| 21:09:35 | zigo | Maybe ... | |
| 21:09:42 | artom | With this dict that gives cell and mq status for all cells and mqs | |
| 21:09:45 | artom | *his | |
| 21:09:47 | zigo | Everyone keeps disturbing me and I can't read IRC properly! :) | |
| 21:10:04 | zigo | Sorry then. | |
| 21:10:25 | artom | No worries :) | |
| 21:16:31 | openstackgerrit | Merged openstack/nova master: Remove future imports https://review.opendev.org/714675 | |
| 23:45:22 | openstackgerrit | Merged openstack/nova master: Follow-up for NUMA live migration functional tests https://review.opendev.org/724903 | |
| 23:57:35 | openstackgerrit | Artom Lifshitz proposed openstack/nova stable/ussuri: Functional tests for NUMA live migration https://review.opendev.org/725736 | |
| #openstack-nova - 2020-05-06 | |||
| 06:34:29 | openstackgerrit | Jie Li proposed openstack/nova master: replace the "hide_hypervisor_id" to "hw:hide_hypervisor_id" https://review.opendev.org/722187 | |
| 06:58:22 | openstackgerrit | Jiri Suchomel proposed openstack/nova master: Add ability to download Glance images into the libvirt image cache via RBD https://review.opendev.org/574301 | |
| 07:29:28 | openstackgerrit | Jie Li proposed openstack/nova master: replace the "hide_hypervisor_id" to "hw:hide_hypervisor_id" https://review.opendev.org/722187 | |
| 07:31:52 | ignaziocassano1 | Hello, I solved live migration issue on EMC iscsi driver for cinder. They did not write multipath.conf needs skip_kpartx yes | |
| 07:32:36 | ignaziocassano1 | Without this parameter when a vm migrate the map in use error is reported | |
| 07:36:18 | bauzas | good morning Nova | |
| 07:40:00 | brinzhang | good morning, bauzas ^^ | |
| 08:32:37 | openstackgerrit | Tony Su proposed openstack/nova-specs master: Re-propose provider-config-file spec for Victoria https://review.opendev.org/725788 | |
| 08:59:24 | aarents | good morning | |
| 09:00:22 | aarents | gibi bauzas I added 2 topics in vptg etherpad, if we have time for this. | |
| 09:03:23 | bauzas | aarents: ack, will look | |
| 09:03:31 | bauzas | I need to do some reviews this week | |
| 09:06:01 | kashyap | sean-k-mooney: stephenfin: Hi, do you recall if Nova enables IOThreads at all? | |
| 09:06:05 | kashyap | I know we don't have a way to configure them | |
| 09:06:11 | stephenfin | no, it doesn't | |
| 09:06:54 | stephenfin | there's a TODO but for now it's emulator threads that we support offloading | |
| 09:07:08 | kashyap | stephenfin: TODO in the code? Got a pointer? /me goes digging in driver.py | |
| 09:08:27 | stephenfin | kashyap: https://github.com/openstack/nova/blob/20.1.0/nova/virt/libvirt/driver.py#L4892-L4907 | |
| 09:09:33 | kashyap | stephenfin: Oh yeah, I recall this bit; thanks! | |
| 09:09:57 | kashyap | stephenfin: This is a low-hanging fruit to improve performance. | |
| 09:12:20 | openstackgerrit | Huaqiang Wang proposed openstack/nova-specs master: Use PCPU and VCPU in one instance https://review.opendev.org/725801 | |
| 09:12:46 | kashyap | stephenfin: I'm going to file a proper upstream tracker / blueprint for it (after checking if one exists) | |
| 09:13:11 | kashyap | Doh, it does exist - https://blueprints.launchpad.net/nova/+spec/libvirt-iothreads | |
| 09:27:26 | kashyap | stephenfin: Oh, looking at Gerrit history, I see there was an attempt to enable IOThreads: https://review.opendev.org/#/c/230968/ | |
| 09:27:36 | kashyap | stephenfin: Err, wrong channel :) | |
| 09:56:26 | openstackgerrit | Lee Yarwood proposed openstack/nova master: compute: Validate a BDMs disk_bus when provided https://review.opendev.org/724842 | |
| 09:58:15 | openstackgerrit | Lee Yarwood proposed openstack/nova master: docs: Add evacuation pre-conditions around the src host https://review.opendev.org/723853 | |
| 10:01:14 | openstackgerrit | Merged openstack/nova master: doc: Fix list rendering in cli/nova-status.rst https://review.opendev.org/724958 | |
| 11:40:07 | openstackgerrit | jayaditya gupta proposed openstack/nova master: Support for --force flag for nova-manage placement heal_allocations command https://review.opendev.org/715395 | |
| 11:56:20 | openstackgerrit | Jie Li proposed openstack/nova master: replace the "hide_hypervisor_id" to "hw:hide_hypervisor_id" https://review.opendev.org/722187 | |
| 12:15:59 | gibi | bauzas: what do you think? Now that RC2 is out should we remove the rc tag from the bug https://bugs.launchpad.net/nova/+bug/1875418 ? | |
| 12:15:59 | openstack | Launchpad bug 1875418 in OpenStack Compute (nova) "Generated policy.json in Ussuri is broken by default" [High,In progress] - Assigned to Ghanshyam Mann (ghanshyammann) | |
| 12:16:13 | bauzas | on a meeting but okay | |
| 12:22:26 | gibi | bauzas: thanks | |
| 13:10:09 | sean-k-mooney | kashyap: no we dont enable iotrheads | |
| 13:10:21 | sean-k-mooney | kashyap: so all io is handeled by the emulator thread | |
| 13:11:17 | sean-k-mooney | kashyap: didnt we decided it was not worth it to enable iothread since we did not want yet another thing to request and the recomendation from the qemu folks was to leave it set to 1 | |
| 13:12:31 | sean-k-mooney | if the instance was pinned and the emulator thread type was shared i could maybe see use spawning 1 io tread per cpu in the shared pool and pinning the io threads to the shared ppol the same way we pinn the emulator thread to float over teh shared pool. | |
| 13:12:43 | sean-k-mooney | or perhaps shared pool -1 io threads | |
| 13:13:23 | sean-k-mooney | but i dont think we should bother enableing them for floating vms and im not sure there would be a significant imporvment for pinned instnaces | |
| 13:31:23 | openstackgerrit | Stephen Finucane proposed openstack/nova master: replace the "hide_hypervisor_id" to "hw:hide_hypervisor_id" https://review.opendev.org/722187 | |
| 13:54:44 | stephenfin | kashyap, sean-k-mooney: Yeah, I was pretty sure danpb or you (kashyap) had said that the benefits weren't justified and emulator thread offloading got us most of the way there | |
| 13:56:16 | francoisp | lyarwood hello, when you have time: https://review.opendev.org/#/c/725272/ (for Rocky this time around) - thanks! | |
| 13:58:42 | kashyap | sean-k-mooney: stephenfin: So this morning, QEMU Block Layer folks pinged me to discuss this | |
| 13:59:07 | kashyap | sean-k-mooney: stephenfin: And they said they're going to publish benchmarks proving that it _is_ a well-worth to enable IOThreads by default | |
| 13:59:30 | kashyap | sean-k-mooney: stephenfin: Let's wait for the documentation / blog post to come, I'll post it here, then we can discuss | |
| 14:00:10 | kashyap | stephenfin: Your recollection is right; but for the past 2 years QEMU has done a tonne of rework in this area. So it worth it to revisit the topic | |