| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2019-02-28 | |||
| 19:52:47 | mriedem | if this code ever lands we should probably have that in the docs somewhere | |
| 19:52:52 | openstackgerrit | Chris Friesen proposed openstack/nova master: Improve libvirt image and snapshot handling https://review.openstack.org/616692 | |
| 19:53:02 | artom | aspiers, how are some COMPUTE_* traits *not* set on RPs? | |
| 19:53:13 | artom | Or did you mean "set on RPs by the admin"? | |
| 19:53:26 | aspiers | artom: if they don't apply? not all traits are set on everything all the time :) | |
| 19:54:04 | artom | Oh, you're using "set" as in actually set, not "capable of being set" | |
| 19:54:09 | aspiers | yes | |
| 19:54:26 | artom | I swear, I dunno what is it about placement | |
| 19:54:28 | aspiers | mriedem: don't scare me with conditional clauses like "if this code ever lands" :-p | |
| 19:54:40 | artom | Every time it's a rabbithole of endless taxonomy and modelling debate | |
| 19:54:56 | melwitt | that diagram is art. and agreed please add it to docs along with patch | |
| 19:55:12 | aspiers | melwitt: where would it belong? | |
| 19:55:19 | artom | aspiers, anyways, yeah, seems to make sense, I'll wait for the code :) | |
| 19:56:27 | melwitt | aspiers: do we have any existing driver capability doc? if so, there. if not, maybe we could make a new doc. and in that case, it could be a follow on patch | |
| 19:56:34 | jaypipes | artom: yes. Nova's NUMA code is much clearer than placement's modeling. | |
| 19:57:23 | aspiers | Now I'm wondering why this diagram attracted so much attention, vs. silence for my other two diagrams in the commit message :) | |
| 19:57:38 | melwitt | would just be nice to capture helpful diagrams like that somewhere | |
| 19:57:50 | aspiers | melwitt: I didn't notice any on a brief scan the other day | |
| 19:57:56 | artom | jaypipes, hah, I'm not saying the code is bad | |
| 19:57:57 | melwitt | well, I happened to click on this one and haven't read your commit message yet | |
| 19:58:07 | aspiers | melwitt: I also made https://pasteboard.co/I25qICd.png | |
| 19:58:17 | artom | jaypipes, I'm observing that placement monopolizes discussions every time it's brought up | |
| 19:58:29 | artom | Even in our RH team call this morning | |
| 19:59:20 | aspiers | melwitt: yeah, I was wondering more about the people who *have* already reviewed it ;-) | |
| 20:00:56 | melwitt | aspiers: heh. well, it's not _that_ important about the diagrams, just a general sentiment of, it would be nice if we had more pictures in our docs for features we have. and since you already made those. another thing to throw on the TODO list | |
| 20:01:35 | aspiers | melwitt: agreed, happy to add to docs. I would have already included in the review if I had found a good place for them | |
| 20:01:51 | jaypipes | aspiers: I haven't reviewed the patch yet. | |
| 20:02:06 | jaypipes | aspiers: been languishing in my endless review queue... | |
| 20:02:07 | efried | aspiers: You still have another rev forthcoming, though, right? | |
| 20:02:10 | jaypipes | apologies. | |
| 20:02:28 | melwitt | aspiers: yeah. don't worry much about it, just another TODO for the pile of neverending TODOs :) | |
| 20:02:46 | aspiers | jaypipes: no worries ;) | |
| 20:02:51 | aspiers | efried: yes, working on it right now | |
| 20:03:15 | efried | cool | |
| 20:03:23 | aspiers | efried: I was planning to fix the issues, then submit, and only then rebase and submit again, to keep rebase changes in a separate patchset | |
| 20:03:26 | aspiers | makes sense? | |
| 20:03:45 | efried | aspiers: ++, hate having to review rebase+updates in same PS. | |
| 20:04:02 | aspiers | efried: one day we will not have to worry about that any more | |
| 20:04:14 | aspiers | o/~ one day ... o/~ | |
| 20:04:18 | efried | yeah, it wasn't long ago that the rebase PS would include all the files from the rebase. | |
| 20:04:30 | efried | ...from all the intervening commits | |
| 20:04:37 | efried | so where we are now is at least an improvement on that. | |
| 20:05:12 | aspiers | well, "long ago" depends very much on whether you are using the timeline of our Gerrit deployments, or the upstream Gerrit releases ;-) | |
| 20:05:31 | aspiers | but yeah, I know you meant the former | |
| 20:05:52 | aspiers | I kinda want to volunteer for the infra team *just* to get us on 2.16 | |
| 20:05:57 | aspiers | but I don't have time :-( | |
| 20:07:21 | aspiers | melwitt: would it go under "Nova Major Subsystems"? https://docs.openstack.org/nova/latest/contributor/index.html | |
| 20:07:52 | aspiers | a new doc called "resource provider trees" or something like that? | |
| 20:12:21 | melwitt | aspiers: yeah, AFAIK, we don't have any nova docs about traits yet. is it only useful to nova contributors or to more ppl? if it's more, then it wouldn't be under contributor/ | |
| 20:12:58 | aspiers | melwitt: I think some would, some wouldn't - depends on whether it's internals | |
| 20:13:31 | aspiers | IMHO this Venn diagram is borderline either way | |
| 20:13:52 | aspiers | I could do a more user-facing version | |
| 20:14:20 | melwitt | no, I think it's ok, was just sanity checking. I think a new subsystem doc would make sense, like you suggested | |
| 20:15:59 | aspiers | melwitt: any ideas about the title? | |
| 20:16:22 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Move legacy-grenade-dsvm-neutron-multinode-live-migration in-tree https://review.openstack.org/640181 | |
| 20:16:23 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Make nova-grenade-live-migration voting and gate https://review.openstack.org/640182 | |
| 20:17:05 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Make nova-grenade-live-migration voting and gating https://review.openstack.org/640182 | |
| 20:17:54 | mriedem | flavors docs talked about traits https://docs.openstack.org/nova/latest/user/flavors.html | |
| 20:18:09 | mriedem | so you'd likely want some reference to how system-driven traits and user-driven traits overlap | |
| 20:18:51 | aspiers | mriedem: good point, thanks | |
| 20:19:13 | mriedem | i'm not sure contributor is really the right place for that information, | |
| 20:19:15 | mriedem | since it is user facing | |
| 20:19:23 | mriedem | user/admin | |
| 20:19:39 | mriedem | i mean we have https://docs.openstack.org/nova/latest/user/#maintenance | |
| 20:19:51 | mriedem | and https://docs.openstack.org/nova/latest/reference/#internals | |
| 20:20:11 | mriedem | the latter is where provider tree is documented | |
| 20:20:15 | mriedem | https://docs.openstack.org/nova/latest/reference/update-provider-tree.html | |
| 20:20:24 | mriedem | so you'd also link from ^ likely | |
| 20:21:12 | mriedem | i'd probably throw something under reference/internals like the upt stuff, and reference from wherever it makes sense, like the flavors doc | |
| 20:21:42 | mriedem | "Note that there are compute-managed traits. Details about those can be found <here>." | |
| 20:22:26 | mriedem | some of those compute traits also link to the feature support matrix ... https://docs.openstack.org/nova/latest/user/support-matrix.html#operation_device_tags | |
| 20:22:39 | mriedem | aspiers: i'd worry about documentation later :) | |
| 20:23:16 | aspiers | yup | |
| 20:44:30 | openstackgerrit | Matt Riedemann proposed openstack/nova stable/rocky: Fix legacy-grenade-dsvm-neutron-multinode-live-migration https://review.openstack.org/640186 | |
| 20:47:15 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Make nova-grenade-live-migration voting and gating https://review.openstack.org/640182 | |
| 20:47:15 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Move legacy-grenade-dsvm-neutron-multinode-live-migration in-tree https://review.openstack.org/640181 | |
| 20:50:16 | melwitt | nova meeting in 10 minutes | |
| 20:51:19 | artom | Oh hey, I can actually this one | |
| 20:51:30 | artom | Usually it's smack in the middle of daycare taxi time | |
| 20:58:30 | openstackgerrit | melanie witt proposed openstack/nova master: Add online data migration for populating user_id https://review.openstack.org/633351 | |
| 20:58:31 | openstackgerrit | melanie witt proposed openstack/nova master: WIP Count instances from mappings and cores/ram from placement https://review.openstack.org/638073 | |
| 20:58:31 | openstackgerrit | melanie witt proposed openstack/nova master: Add get_counts() to InstanceMappingList https://review.openstack.org/638072 | |
| 20:58:32 | openstackgerrit | melanie witt proposed openstack/nova master: Use instance mappings to count server group members https://review.openstack.org/638324 | |
| 20:59:56 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Make nova-grenade-live-migration voting and gating https://review.openstack.org/640182 | |
| 20:59:56 | openstackgerrit | Matt Riedemann proposed openstack/nova master: Move legacy-grenade-dsvm-neutron-multinode-live-migration in-tree https://review.openstack.org/640181 | |
| 21:09:42 | openstackgerrit | Matt Riedemann proposed openstack/nova stable/rocky: Move legacy-grenade-dsvm-neutron-multinode-live-migration in-tree https://review.openstack.org/640191 | |
| 21:41:19 | openstackgerrit | Matt Riedemann proposed openstack/nova stable/queens: Fix legacy-grenade-dsvm-neutron-multinode-live-migration https://review.openstack.org/640197 | |
| 21:42:16 | openstackgerrit | Matt Riedemann proposed openstack/nova stable/queens: Move legacy-grenade-dsvm-neutron-multinode-live-migration in-tree https://review.openstack.org/640198 | |
| 21:49:52 | marst | Hello. I've installed devstack (with minimal local.conf), how do I create several fake hypervisors? I remember there was a blog post from someone with 1000 hypervisors, but can't find it. | |
| 21:50:34 | marst | Just trying to teach people how nova-scheduler filters work | |
| 21:50:49 | marst | (but first need to understand it myself) | |
| 21:51:27 | mriedem | marst: this would get you partway https://review.openstack.org/#/c/507918/ | |
| 21:52:10 | mriedem | the main things are (1) use the fake virt driver (2) noop quota driver and (3) $NUMBER_FAKE_NOVA_COMPUTE | |
| 21:52:41 | mriedem | even with that if you create too many fake computes you can run out of resources on a single host | |
| 21:52:53 | marst | mriedem: Thanks a lot!! This is perfect! | |
| 21:53:17 | mriedem | in fact i think i changed devstack so that if you use the fake virt driver, you get the noop quota driver automatically | |
| 21:53:43 | mriedem | https://review.openstack.org/#/c/571933/ | |
| 21:54:09 | mriedem | yeah so you just need to say VIRT_DRIVER=fake and NUMBER_FAKE_NOVA_COMPUTE=whatever | |
| 21:54:21 | marst | mriedem: awesome! thank you! | |
| 21:54:25 | mriedem | nop | |