*** wxy-xiyuan has joined #openstack-glance | 01:17 | |
*** MattMan_1 has quit IRC | 01:43 | |
*** MattMan_1 has joined #openstack-glance | 01:43 | |
*** mhen has quit IRC | 01:58 | |
*** mhen has joined #openstack-glance | 02:00 | |
*** LiangFang has joined #openstack-glance | 03:24 | |
*** pooja_jadhav has joined #openstack-glance | 04:33 | |
*** abhishekk has joined #openstack-glance | 04:38 | |
*** ratailor has joined #openstack-glance | 05:53 | |
*** ratailor has quit IRC | 06:28 | |
*** mosulica has joined #openstack-glance | 06:31 | |
*** ratailor has joined #openstack-glance | 06:35 | |
*** rcernin has quit IRC | 07:21 | |
*** belmoreira has joined #openstack-glance | 08:17 | |
openstackgerrit | Merged openstack/glance master: Handle StopIteration for Py3.7 PEP 0479 https://review.openstack.org/582628 | 08:36 |
---|---|---|
*** LiangFang has quit IRC | 10:05 | |
*** abhishekk has quit IRC | 10:15 | |
*** rcernin has joined #openstack-glance | 10:35 | |
*** dpawlik has joined #openstack-glance | 10:37 | |
*** Luzi has joined #openstack-glance | 10:46 | |
*** rcernin has quit IRC | 11:17 | |
*** ratailor has quit IRC | 11:20 | |
*** belmorei_ has joined #openstack-glance | 11:48 | |
*** belmoreira has quit IRC | 11:51 | |
*** mosulica has quit IRC | 12:02 | |
*** mosulica has joined #openstack-glance | 12:27 | |
*** brinzhang has quit IRC | 12:28 | |
*** zul has quit IRC | 13:38 | |
*** kukacz has quit IRC | 13:54 | |
*** kukacz has joined #openstack-glance | 13:54 | |
*** munimeha1 has joined #openstack-glance | 14:08 | |
*** rosmaita has joined #openstack-glance | 14:13 | |
*** itlinux has quit IRC | 14:50 | |
*** gcb_ has joined #openstack-glance | 14:53 | |
*** mosulica has quit IRC | 15:04 | |
*** Luzi has quit IRC | 15:17 | |
*** itlinux has joined #openstack-glance | 15:43 | |
aspiers | hi, please can someone point me at docs for how OS::Compute::HostCapabilities can be used? | 16:01 |
aspiers | It claims to use ComputeCapabilitiesFilter, but isn't it achieving pretty much the same thing as ImagePropertiesFilter? When should one be used over the other? | 16:01 |
*** gcb_ has quit IRC | 16:03 | |
rosmaita | aspiers: I guess you've looked at https://docs.openstack.org/nova/latest/admin/configuration/schedulers.html ? I think this is really a nova question | 16:08 |
aspiers | rosmaita: yeah, I've looked at that, but this is about image metadata | 16:08 |
rosmaita | have you seen https://docs.openstack.org/glance/latest/admin/useful-image-properties.html ? | 16:09 |
aspiers | it's defined in https://github.com/openstack/glance/blob/master/etc/metadefs/compute-host-capabilities.json | 16:09 |
aspiers | yes I've seen that too | 16:09 |
rosmaita | aspiers: the metadefs by themselves don't do anything, they just give you a programmatic way to find namespaces and values | 16:10 |
rosmaita | you have to actually put the info on an image as additional image properties in order to get something to happen | 16:10 |
aspiers | right, so that schema is associating the capabilities: prefix with flavors, right? | 16:10 |
*** belmorei_ has quit IRC | 16:11 | |
aspiers | I already knew about ComputeCapabilitiesFilter, but I thought the idea was that capabilities: extra specs could also be put on glance images? | 16:12 |
aspiers | otherwise why is this file in the glance project? | 16:12 |
rosmaita | aspiers: the metadefs are in Glance ... for no particular reason | 16:12 |
aspiers | or is it because nova's extra spec mechanism relies on namespaces/schema which are defined by glance? | 16:13 |
aspiers | ah | 16:13 |
aspiers | so the only way to require a particular CPU capability on an image is via the new trait:... feature? | 16:13 |
rosmaita | yeah, Glance is just providing a home for the metadata definitions catalog | 16:13 |
aspiers | OK | 16:13 |
rosmaita | well, there are 2 steps, i think | 16:13 |
aspiers | thank you! that sheds a lot of light where I was previously confused | 16:13 |
rosmaita | you need to have nova configured properly | 16:13 |
rosmaita | and then you need to put the metadata on images that require it | 16:14 |
rosmaita | (don't know if that helps) | 16:14 |
aspiers | sure, you need the nova filter configured | 16:14 |
aspiers | but IIUC, if I use traits then there is no need for that | 16:14 |
aspiers | trait filtering seems to be built-in, both on images and flavors | 16:14 |
aspiers | AFAICS | 16:14 |
rosmaita | i think you are correct | 16:16 |
aspiers | I have to dash now but this was extremely helpful - thanks a LOT! | 16:16 |
rosmaita | the basic deal is that you can run a cloud without putting special stuff on the images; it's only when your infrastructure is complicated (for example, mixed hypervisors) that it may be useful to put stuff on the images | 16:17 |
rosmaita | aspiers: np, hth | 16:17 |
imacdonn | rosmaita: I added a little bit more to the etherpad since Friday .... looks like no one else has :/ Actually, though, at the end of the day, it's basically what I had already implemented, with some clarification on the schema addition | 16:29 |
rosmaita | imacdonn: ok, thanks, i'll take a look after lunch. i'll also make sure it gets discussed at the thursday meeting | 16:29 |
rosmaita | which, i know, is a long ways away from your point of view! | 16:30 |
imacdonn | rosmaita: I'd like to know what you need to see added/changed in the spec-lite, so at least we can hopefully get that finally approved ... also if you have any comments on the implementation, that's up for review too. Thanks! | 16:31 |
rosmaita | sounds good, i'll get you some comments later today | 16:31 |
imacdonn | cool | 16:31 |
*** gyee has joined #openstack-glance | 16:48 | |
prometheanfire | so, now that it's monday... | 16:50 |
prometheanfire | new oslo-messaging seems to be breaking things http://logs.openstack.org/21/607521/2/check/cross-glance-py35/e2161d7/testr_results.html.gz | 16:50 |
*** imacdonn has quit IRC | 18:23 | |
*** imacdonn has joined #openstack-glance | 18:23 | |
openstackgerrit | Brian Rosmaita proposed openstack/glance master: Do not use oslo.messaging 9.0.0 https://review.openstack.org/608760 | 20:06 |
rosmaita | prometheanfire: ^^ | 20:06 |
imacdonn | that seems like the ostrich approach? Will have to deal with the deprecation at some point....... | 20:15 |
rosmaita | from the patch to oslo.messaging master, looks like the problem is a missing default value on the oslo side | 20:21 |
imacdonn | that may be one problem, but there's also a deprecated option that's been removed | 20:22 |
rosmaita | feel free to -1 the patch, i did not look too closely into this other than to note that our tests all pass against oslo.messaging master | 20:23 |
imacdonn | hmm, strang | 20:24 |
imacdonn | I haven't gotten to try yet ... but I do see use of the removed option (rpc_backend) in there, so not sure how it could work | 20:25 |
rosmaita | yeah, i may be over-relying on our test coverage | 20:25 |
imacdonn | how did you test with oslo.messaging master ? | 20:25 |
rosmaita | There | 20:31 |
rosmaita | 's a periodic job defined that tests all the oslo project masters against glance master | 20:32 |
rosmaita | for unit and functional tests | 20:32 |
rosmaita | if you want to do it locally, i'm not sure how to do it short of hacking the patch into your tox virtualenvs | 20:33 |
imacdonn | for cinder, I activated the venv, and used pip to upgrade oslo.messaging (to 9.0.0) .. that reproduced the problems | 20:34 |
imacdonn | activated the tox py27 venv, that is | 20:35 |
rosmaita | i think i have paste somewhere from when i had to do this with glance_store master, hang on, i''ll look | 20:35 |
rosmaita | imacdonn: http://paste.openstack.org/show/726942/ | 20:37 |
imacdonn | rosmaita: OK, yeah, I guess that would work for testing a patch that's not merged yet | 20:42 |
rosmaita | there's probably a fancier way to do it, but this works if you just need to test one change that isn't available in a release yet | 20:44 |
imacdonn | right | 20:44 |
imacdonn | I'd want to start by reproducing the problem - I may have a go this afternoon ... working on a cinder patch right now | 20:45 |
rosmaita | imacdonn: if you have time, that would be great | 20:45 |
*** pcaruana has quit IRC | 20:49 | |
openstackgerrit | Brian Rosmaita proposed openstack/glance master: Clarify the Glance/metadefs relationship https://review.openstack.org/608769 | 20:53 |
*** prometheanfire has left #openstack-glance | 20:59 | |
*** itlinux has quit IRC | 21:08 | |
imacdonn | rosmaita: on closer inspection, the only mentions of rpc_backend in the glance source are in the example config files - e.g. https://github.com/openstack/glance/blob/master/etc/glance-api.conf#L1841-L1846 | 21:44 |
imacdonn | whilst that won't break any tests, it prob should be removed | 21:45 |
rosmaita | good to know, thanks for checking | 21:56 |
rosmaita | imacdonn: that should be corrected automatically when the config file is regenerated | 21:57 |
rosmaita | so nothing to do ... the best kind of change! | 21:57 |
imacdonn | I was going to try to hack in the oslo.messaging fix and see if it makes all the other problems go away | 21:59 |
rosmaita | :) | 21:59 |
imacdonn | and apparently it does! | 22:02 |
imacdonn | where does that example config auto-generation happen ? | 22:04 |
rosmaita | it has to be triggered manually | 22:04 |
imacdonn | to it'll get manually automatically corrected? ;) | 22:05 |
rosmaita | imacdonn: https://docs.openstack.org/glance/latest/contributor/refreshing-configs.html | 22:05 |
rosmaita | yeah, i know | 22:05 |
imacdonn | heh | 22:05 |
*** munimeha1 has quit IRC | 22:50 | |
*** rcernin has joined #openstack-glance | 22:50 | |
*** gyee has quit IRC | 23:57 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!