Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2018-09-04
14:49:27 sean-k-mooney tobias-urdin: this kind of think is something that could be a feature request to the openstacksdk/shade teams as getting the root volume for boot from volume is likely one of thoes things that will change alot depending on your env and could be handeld in a shade proxy api
14:49:46 mdbooth tobias-urdin: In practise, it's probably going to be the one with volume_image_metadata.
14:50:28 mdbooth Although I was reviewing a v2v tool the other day which added volume_image_metadata to multiple volumes, which would have broken that assumption, but I can't imagine that's common.
14:51:20 mdbooth Creation data is less reliable, especially if you have persistent data on volume X, and created a root volume to contain an app to manipulate it some time later.
14:51:47 mdbooth Unlike the first case, I'd expect that to be likely to happen in the wild.
14:52:39 tobias-urdin sean-k-mooney: sort of pressing to get this stuff done (as always...) otherwise it would've been optimal to check the sdk first
14:53:04 tobias-urdin mdbooth: yeah, i was thinking about combining if there is multiple volume_image_metadata just assume the first created.. atleast i would always get something
14:56:03 prometheanfire stephenfin: hi :D
14:56:09 stephenfin prometheanfire: o/
14:57:16 mdbooth Not strictly related, but afaict we attach volumes in a non-deterministic order on restart, except for the root volume. And that's ok. We should probably deliberately randomise it :)
14:58:39 tobias-urdin while I'm at it... here's a sad question, can I somehow block the creation of image backed instances? (i.e images_type backed instances)
14:59:29 tobias-urdin policy, super simple api hack or similar, everything's allowed but the best would be to not really touch anything critical
14:59:38 prometheanfire stephenfin: maybe my deployment is just broken, gimme a few to get out of this meeting and I'll update you
14:59:59 tobias-urdin I was hoping on setting images_type to None but that errors out upon initialization
15:01:25 mdbooth tobias-urdin: Don't deploy glance?
15:01:33 tobias-urdin (oh how I wish there was a cinder backend for images_type right now, wish i familar enough with the codebase to drive such a work)
15:02:41 tobias-urdin mdbooth: how u mean? hm settings [glance]/api_servers to something invalid?
15:02:58 sean-k-mooney tobias-urdin: well if you have ceph you could set nova to image_type ceph
15:03:38 tobias-urdin sean-k-mooney: yea that's what we do today, however don't want users that way at all, weird i know but we want to default to cinder
15:03:46 efried bauzas, mriedem: https://review.openstack.org/#/c/598365/ just needs config helps updated?
15:04:14 mriedem haven't looked at reviews on it yet today
15:04:39 sean-k-mooney tobias-urdin: its not that weird haveing a generic image_type cinder has come up in dublin ptg and last time in denver
15:04:42 bauzas efried: yep, IMHO
15:04:50 sean-k-mooney its just not that simple.
15:05:05 bauzas efried: mriedem: once done, we could modify the default values
15:05:26 mriedem i thought ovh already had a patch for that
15:05:29 tobias-urdin sean-k-mooney: yea :(
15:05:35 efried bauzas, mriedem: It's blocking our CI, so I would like to get it merged as soon as possible.
15:05:48 bauzas orly N
15:05:49 bauzas ?
15:05:55 bauzas if so, let me +W it
15:06:03 mriedem bauzas: https://review.openstack.org/#/c/532924/
15:06:10 sean-k-mooney mriedem: do you know if anyone is proposing or looking at a cinder image time for stein or makeing bfv the default?
15:06:19 efried bauzas: Okay, thanks. We can do the conf helps in a fup?
15:06:30 sean-k-mooney *image_type=cinder
15:06:31 mriedem efried: i can look and update
15:06:44 bauzas efried: if needed, yep
15:06:51 mriedem sean-k-mooney: as in the fabled libvirt cinder image backend of lore?
15:07:00 mriedem no, no one is working on that
15:07:14 bauzas mriedem: would you be possible to pass a new revision now, or just a new change later ?
15:07:18 bauzas if the latter, no worries
15:07:24 kashyap sean-k-mooney: When you get a minute, please remind me again: we can't set PCIe root ports via flavor metadata property, can we?
15:07:35 sean-k-mooney mriedem: ya that is what i had rememberd form dublin
15:07:49 mriedem bauzas: efried: i'll update it in a minute
15:07:53 bauzas mriedem: and yeah, I remember this change
15:07:59 prometheanfire stephenfin: ya, I made sure no pycache/pyc/pyo
15:08:24 sean-k-mooney kashyap: if we can the glance metadef have not been created to document it. i wish we could and didn not have this in the nova config. ill check
15:08:56 kashyap sean-k-mooney: Right, I presume we _can't_ today; I'll go look the code
15:09:02 bauzas mriedem: and I also remember the spec https://review.openstack.org/#/c/552105/3/specs/rocky/approved/default-allocation-ratios.rst
15:09:10 stephenfin prometheanfire: There's definitely some form of caching going on or your source is located somewhere else. That's the only reason for that stuff to happen
15:09:15 bauzas mriedem: but I think it's a separate issue
15:09:20 kashyap sean-k-mooney: A libvirt dev was asking that question: can Nova set the root ports via flavor; or just through a global knob
15:10:46 prometheanfire stephenfin: even odder, I ran the py27 version of nova-compute, got the traceback containing py35
15:10:59 prometheanfire the main traceback was py27 though
15:11:12 stephenfin prometheanfire: Can you paste the output of that?
15:11:17 prometheanfire I think the real issue was that portion
15:12:20 sean-k-mooney kashyap: this is the config generation code but github is not finding its usage https://github.com/openstack/nova/blob/c6218428e9b29a2c52808ec7d27b4b21aadc0299/nova/virt/libvirt/config.py#L1713-L1728
15:13:01 sean-k-mooney kashyap: codesearch did however http://git.openstack.org/cgit/openstack/nova/tree/nova/virt/libvirt/driver.py#n5117
15:13:21 sean-k-mooney kashyap: the dirver uses the conf value directly
15:13:37 kashyap sean-k-mooney: So no metadata property
15:13:43 kashyap Wonder if we should file a blueprint to add it
15:14:07 sean-k-mooney kashyap: i would be +1 on that espcially if we can deprecate and remvoe the conf option
15:14:27 kashyap sean-k-mooney: Yeah, we should be able to do that; instead of the global config.
15:14:45 kashyap Damn, since I 'discovered' the bug, I get the pleasure of filing the Blueprint I guess :P
15:14:58 kashyap sean-k-mooney: More seriously, paperwork question: does this require a spec? Since it's user-impacting?
15:15:29 kashyap Or a spec-less blueprint is reasonable enough?
15:16:41 sean-k-mooney kashyap: spec-less blueprint. extra specs are a gray area as they are not technical part of the versioned api but are user fasing so mriedem or someelse will likely comment on the blueprint if a specs is needed
15:17:18 prometheanfire stephenfin: https://gist.githubusercontent.com/prometheanfire/6512134e799ec8c08c3f080150f60d19/raw/7cdb3a9a350c19540b1c930077e164786226636b/gistfile1.txt
15:17:20 kashyap sean-k-mooney: Yep, noted; thanks for the discussion.
15:18:25 sean-k-mooney by the way i have been using http://codesearch.openstack.org/ a lot more recently instead of using github to such for these things its pretty good
15:19:07 kashyap Ah, nice.
15:19:23 kashyap sean-k-mooney: BTW, seems like "hw:machine_type" isn't documented a flavor extra specs here: https://docs.openstack.org/nova/latest/user/flavors.html
15:19:33 kashyap Is that so?
15:20:01 sean-k-mooney kashyap: correct it is documented here https://github.com/openstack/glance/blob/master/etc/metadefs/compute-libvirt-image.json#L34-L38
15:20:44 sean-k-mooney accouding to the glace metadef registry its only valid on image not flavor
15:21:22 kashyap sean-k-mooney: Hmm, but the syntax is slightly different: hw:machine_type vs hw_machine_type (in nova.conf)
15:21:37 kashyap Right, this seem to work: `openstack flavor set --property hw:machine_type=x86_64=q35 test.q35`
15:21:41 stephenfin prometheanfire: Based on that, it seems oslo.service package in your virtualenv is starting a thread using the system oslo.service package. I've no idea why that would happen
15:21:59 stephenfin prometheanfire: Might be worth asking on #openstack-oslo to see if anyone else has seen this before
15:22:11 sean-k-mooney kashyap: hw: is the namesapced flavor syntax, images dont have namespaces so the namespace is prepended with an _ instead of :
15:22:40 sean-k-mooney kashyap: https://github.com/openstack/glance/blob/master/etc/metadefs/compute-libvirt-image.json#L7-L11 tell you these are only valid in the image
15:22:44 stephenfin prometheanfire: You've patched 'usr/lib64/python3.5/site-packages/nova', I assume?
15:22:49 kashyap sean-k-mooney: Nod; I'll send a doc patch to document it here as well: https://docs.openstack.org/nova/latest/user/flavors.html
15:23:23 sean-k-mooney kashyap: if its valid in both it looks like https://github.com/openstack/glance/blob/master/etc/metadefs/compute-libvirt.json#L7-L16
15:23:25 prometheanfire stephenfin: that file didn't exist at the time :P
15:23:39 prometheanfire I uninstalled nova/oslo-service/oslo-db system-wide
15:23:42 prometheanfire only available in the venv
15:23:48 sean-k-mooney kashyap: documenting it there is fine but the autritive source is glance
15:23:49 prometheanfire like I said, broken :P
15:24:13 stephenfin prometheanfire: Very :)
15:24:24 kashyap sean-k-mooney: Yep, noted. (And it seems to be valid for both, IIUC)
15:24:44 stephenfin prometheanfire: Yeah, I'm not sure how much I can help with that. There's something funky going on with venvs that I don't understand. I don't think it's anything to do with the patch itself
15:27:16 prometheanfire stephenfin: ya, at this point justmergeit
15:27:34 sean-k-mooney kashyap: doing a code search i only see code for using it from the conf or the image not the flavor
15:27:51 sean-k-mooney kashyap: http://codesearch.openstack.org/?q=machine_type&i=nope&files=&repos=nova
15:28:12 kashyap sean-k-mooney: Yeah, I've just done a test, indeed it's so
15:28:24 kashyap stephenfin: Hey, when you get a moment, yesterday you said this worked for you:
15:28:42 kashyap $ openstack flavor set --property hw:machine_type=x86_64=q35 test.q35
15:28:42 kashyap $ openstack flavor create test.q35
15:28:45 kashyap $ openstack server create --flavor test.q35 --image test \

Earlier   Later