| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2017-09-08 | |||
| 13:29:00 | stephenfin | Excellent, thanks gibi :) | |
| 13:37:09 | mriedem | johnthetubaguy: policy question on scope checking, | |
| 13:37:24 | mriedem | this is something that continues to confuse me about whether or not non-admins can see extra specs for a flavor | |
| 13:37:46 | mriedem | default policy is admin or owner | |
| 13:37:46 | mriedem | "os_compute_api:os-flavor-extra-specs:index": "rule:admin_or_owner" | |
| 13:38:09 | mriedem | but when we check the scope, we don't pass the flavor in, and the flavor doesn't have a user_id/project_id (owner) anyway | |
| 13:38:38 | mriedem | https://github.com/openstack/nova/blob/master/nova/api/openstack/compute/flavors_extraspecs.py#L53 | |
| 13:38:57 | mriedem | so it's unclear to me if a non-admin can by default view flavor extra specs | |
| 13:43:01 | johnthetubaguy | mriedem: just reading back your note | |
| 13:44:24 | johnthetubaguy | mriedem: we have a default target of {project_id: context.project_id, user_id:context.user_id}, basically making the scope check a no op | |
| 13:45:15 | johnthetubaguy | mriedem: well my no-op I mean always returns true | |
| 13:45:21 | johnthetubaguy | s/my/by/ | |
| 13:45:41 | mriedem | ok so basically anyone can view flavor extra specs by default | |
| 13:47:21 | johnthetubaguy | mriedem: it seems that way, I am just checking the DB code | |
| 13:48:42 | johnthetubaguy | mriedem: it looks like we join extra_specs for everyone here: https://github.com/openstack/nova/blob/master/nova/db/sqlalchemy/api.py#L5017 | |
| 13:49:15 | mriedem | yup | |
| 13:49:26 | mriedem | well, if the flavor is in a cell still, but yeah | |
| 13:49:44 | mriedem | https://github.com/openstack/nova/blob/master/nova/objects/flavor.py#L295 is the new hotness | |
| 13:50:23 | johnthetubaguy | mriedem: oh right, we only hit the other for the unmigrated (probably no one) case | |
| 13:50:52 | mriedem | yeah we can remove that old compat now https://review.openstack.org/#/c/460377/ | |
| 13:50:58 | mriedem | but anyway, that answers my question | |
| 13:51:08 | johnthetubaguy | I see what you did there | |
| 13:51:12 | mriedem | :) | |
| 13:51:18 | mriedem | it was not subtle | |
| 13:51:27 | johnthetubaguy | effective though | |
| 13:51:38 | mriedem | i'm getting some questions about extra specs vs scheduler hints and thinking about things like what is discoverable etc | |
| 13:52:12 | mriedem | and now that i think about it, we really have no discoverable way for an end user to know what scheduler hints are available in a nova installation do we | |
| 13:52:22 | johnthetubaguy | I know sfinucan once expressed and interest doing an image meta like thing for extra specs | |
| 13:52:28 | mriedem | i assume we have just always relied on clouds to document what they support for scheduler hints | |
| 13:52:29 | openstackgerrit | Merged openstack/osc-placement master: tests: add a hook for functional testing in the gate https://review.openstack.org/452122 | |
| 13:53:12 | johnthetubaguy | yeah, we did talk about getting alaski to look at a way to discover available scheduler hints, so we could do API validation on them | |
| 13:53:27 | mriedem | we == rackspace? | |
| 13:53:28 | johnthetubaguy | like have filters and weights register what they support, etc | |
| 13:53:41 | johnthetubaguy | we as in upstream years ago | |
| 13:53:49 | johnthetubaguy | in some summit session | |
| 13:53:50 | mriedem | ok | |
| 13:53:52 | mriedem | we have https://github.com/openstack/nova/blob/master/nova/api/openstack/compute/schemas/scheduler_hints.py | |
| 13:54:03 | mriedem | but the hints in there might not be enabled in the filters, | |
| 13:54:04 | mikal | Do I really have to get up? Or can I just go back to bed? | |
| 13:54:06 | mriedem | and you can load your own | |
| 13:54:17 | mriedem | mikal: just work from bed | |
| 13:54:17 | johnthetubaguy | https://github.com/openstack/nova/blob/master/nova/api/openstack/compute/schemas/scheduler_hints.py#L67 | |
| 13:54:19 | johnthetubaguy | yeah | |
| 13:54:20 | mriedem | like a monarch | |
| 13:54:31 | mriedem | king mikal | |
| 13:55:20 | mikal | mriedem: I can't. I have to go to a big room in the urine district and call people liars | |
| 13:55:39 | mikal | mriedem: its a community service that I offer | |
| 13:56:01 | mriedem | so you're going to go yell at people in a public bathroom? | |
| 13:56:10 | mriedem | i'm unclear on what australians refer to as "the urine district" | |
| 13:56:23 | mikal | mriedem: the urine district is a region of San Francisco | |
| 13:56:29 | mikal | mriedem: most of the city as best as I can tell | |
| 13:56:29 | mriedem | oh i see | |
| 13:56:44 | mriedem | that's less interesting | |
| 13:57:02 | johnthetubaguy | I remember a smell there, but I just hoped it was chlorine in the water with lots of water sprinkler or something | |
| 13:58:15 | mikal | johnthetubaguy: ummm, no | |
| 13:58:29 | mikal | johnthetubaguy: also, don't ask why the BART escalators don't work, and definitely don't touch them | |
| 13:58:42 | johnthetubaguy | mikal: ...true | |
| 13:58:45 | stephenfin | Yay, johnthetubaguy is back | |
| 13:58:59 | mikal | johnthetubaguy: http://www.sfgate.com/bayarea/article/Human-waste-shuts-down-BART-escalators-3735981.php | |
| 13:59:04 | johnthetubaguy | yeah, I am around, now an then :) | |
| 14:01:06 | gibi | bauwser: you were +2 on this patch before, do you still like it? https://review.openstack.org/#/c/498596 | |
| 14:01:12 | sdague | johnthetubaguy: the discoverable bit there was kicked around | |
| 14:01:22 | sdague | but it was all a concept conversation | |
| 14:01:54 | bauwser | gibi: sure, I just have a shit of paperwork coming in litterally by the day before I'm traveling to the PTG, yay \o/ | |
| 14:02:21 | johnthetubaguy | sdague: that sounds right | |
| 14:03:03 | gibi | bauwser: sorry to hear that. Don't worry about that patch now. I can ping you during the PTG | |
| 14:03:18 | bauwser | gibi: no, I just briefly reviewed it, +Wd | |
| 14:03:32 | bauwser | the modifications weren't that huge | |
| 14:03:44 | boolman | when booting a nova instance and not specifying the volume, should nova then create a volume based on the flavor and image_type ? | |
| 14:05:01 | mriedem | gdi our docs search is terrible now | |
| 14:06:00 | mriedem | searching for "flavor extra specs" in docs.o.o mostly gives you forum questions now | |
| 14:06:16 | johnthetubaguy | eek | |
| 14:06:23 | gibi | bauwser: thanks! | |
| 14:06:47 | bauwser | mriedem: you already pointed that out in a ML thread, right? | |
| 14:06:49 | bauwser | no answer yet ? | |
| 14:06:56 | bauwser | (AFAICR) | |
| 14:07:40 | mriedem | bauwser: yes, and sdague started something to define an in-scope search vs a global search | |
| 14:07:49 | mriedem | in-scope meaning, within *these* docs | |
| 14:07:49 | bauwser | k | |
| 14:07:51 | mriedem | like the nova docs | |
| 14:07:54 | johnthetubaguy | boolman: I think you want #openstack, but look at https://developer.openstack.org/api-ref/compute/#create-server and "block_device_mapping_v2.destination_type" basically we only create volumes when requested directly, AFAIK | |
| 14:07:59 | bauwser | yeah, looks a good idea :) | |
| 14:08:18 | sdague | mriedem: that was merged, I don't know if it was released yet | |
| 14:09:10 | boolman | johnthetubaguy: ye sry, I read #topic after I posted. | |
| 14:09:21 | johnthetubaguy | boolman: no worries | |
| 14:10:10 | openstackgerrit | Merged openstack/nova master: Set regex flag on ostestr command for osprofiler tests https://review.openstack.org/501842 | |
| 14:13:58 | mriedem | there are two things wrong with the note at the top of this page https://docs.openstack.org/nova/latest/admin/flavors.html | |
| 14:14:09 | mriedem | 1. wrong policy rule now | |
| 14:14:16 | mriedem | 2. the horizon thing was changed in pike i think | |
| 14:14:55 | mriedem | robcresswell: ^ for "The Dashboard simulates the ability to modify a flavor by deleting an existing flavor and creating a new one with the same name." what exactly does horizon do now? just warn you? or does it not allow you to 'edit' a flavor that is tied to an instance? | |
| 14:15:13 | mriedem | ah https://docs.openstack.org/releasenotes/horizon/pike.html#upgrade-notes | |
| 14:15:17 | johnthetubaguy | mriedem: wow, there is a lot more documented there than I expected to see | |
| 14:15:19 | mriedem | "The ability to edit flavors is disabled by default. See ENABLE_FLAVOR_EDIT in the settings documentation for more information." | |
| 14:17:27 | stephenfin | mriedem: Seeing as you're looking at that file, want to review https://review.openstack.org/#/c/501342/2 | |
| 14:17:38 | sdague | TheJulia / dtantsur: this working as designed - https://bugs.launchpad.net/nova/+bug/1715646 - or a real bug | |
| 14:17:39 | openstack | Launchpad bug 1715646 in OpenStack Compute (nova) "Compute host for ironic is not mapped to a cell if no ironic node is created" [Undecided,New] | |
| 14:17:43 | stephenfin | This is the first step. I'm happy to clean that item above as a follow up | |
| 14:18:15 | dtantsur | sdague: it's a bug, and I thought dansmith fixed it some time ago.. | |
| 14:18:34 | stephenfin | johnthetubaguy: Unfortunately a lot of it is wrong :( especially the stuff around extra_specs | |
| 14:18:49 | johnthetubaguy | stephenfin: joys | |