Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2017-01-03
15:09:12 mriedem i have hot water again! \o/
15:09:16 jaypipes not sure really :(
15:09:16 satyar sure
15:09:21 satyar checking
15:10:43 jaypipes mriedem: well played.
15:10:51 mriedem satyar: you have to also have cinder configured for rbd auth if you're doing boot from volume and expect that encryption
15:11:13 mriedem satyar: in ocata, nova takes the values from the cinder connection and uses those for the auth config in the guest,
15:11:28 mriedem whereas before ocata nova would fall back to it's own local rbd auth config for those values if cinder wasn't configured
15:11:32 mriedem which was a bug
15:11:44 mriedem jaypipes: if i can get my dryer fixed i'll be in heaven
15:11:53 mriedem satyar: http://docs.openstack.org/releasenotes/nova/unreleased.html
15:11:58 mriedem When making connections to Ceph-backed volumes via the Libvirt driver, the auth values (rbd_user, rbd_secret_uuid) are now pulled from the backing cinder.conf rather than nova.conf. The nova.conf values are only used if set and the cinder.conf values are not set, but this fallback support is considered accidental and will be removed in the Nova 16.0.0 Pike release. See the Ceph documentation for configuring Cinder for RBD aut
15:12:21 mriedem which points you here docs.ceph.com/docs/master/rbd/rbd-openstack/#configuring-cinder
15:13:06 satyar mriedme that means the cinder.conf will have the rbd_secret_uuid?
15:13:25 jaypipes satyar: yes, looks like it.
15:13:34 satyar hmmm
15:13:40 cdent rfolco: so a couple of issues: running one test with subunit won't work because the way test generation and discovery works, but you can get at just one from testr with: tox -efunctional test_placement_api.inventory_inventories_for_missing_provider
15:13:53 satyar now i have to change the things :)
15:14:08 satyar Thanks a lot jaypipes and mriedem
15:14:12 mriedem yw
15:14:17 satyar will update how it goes
15:14:24 cdent rfolco: however you can't really use the debugger in that setup because the stdin/out of the debugger is on the wrong side of the wsgi-intercept running the placement application
15:14:27 jaypipes satyar: not sure I did much, but you're welcome anyway :)
15:14:53 cdent rfolco: and because of the stdout capturing that is happening. if you can work around that, it might be doable.
15:14:59 rfolco cdent, that describes exactly what my problem is :)
15:15:01 cdent I tend to just sys.stderr.write all over the place :)
15:15:33 rfolco good advice cdent thx
15:15:48 cdent if you turn off the stdout/in capturing fixture you might be able to make some headway, but I'm not sure
15:16:34 cdent rfolco: i need to run off for a little while but will be back in about an hour, let me know how it goes and if I can be of more help
15:16:56 openstackgerrit Samantha Blanco proposed openstack/nova: Add support matrix for attach and detach interfaces https://review.openstack.org/409967
15:26:20 openstackgerrit Jay Pipes proposed openstack/nova: placement: RT now adds proper Ironic inventory https://review.openstack.org/404472
15:26:21 openstackgerrit Jay Pipes proposed openstack/nova: rt: explicitly pass compute node to _update() https://review.openstack.org/398472
15:26:23 openstackgerrit Jay Pipes proposed openstack/nova: rt: use a single ResourceTracker object instance https://review.openstack.org/398473
15:31:12 openstackgerrit Julian proposed openstack/nova: Create schema generation for NetworkModel https://review.openstack.org/408737
15:32:58 openstackgerrit Samantha Blanco proposed openstack/nova: Add support matrix for attach and detach interfaces https://review.openstack.org/409967
15:34:16 dansmith jaypipes: did you ever return to this to read my comment? https://review.openstack.org/#/c/407238/
15:34:41 jaypipes dansmith: hey Dan, yeah actually I just had that up...
15:35:00 openstackgerrit Rafael Folco proposed openstack/nova-specs: DELETE all inventories for a resource provider https://review.openstack.org/415885
15:35:01 jaypipes dansmith: I'm cool fixing just that area.
15:39:25 openstackgerrit Rafael Folco proposed openstack/nova-specs: DELETE all inventories for a resource provider https://review.openstack.org/415885
15:59:42 kashyap sahid (or anyone interested in this area): Hey there, when you have a moment, mind looking at this -- https://review.openstack.org/#/c/411936/ (libvirt: avoid generating script with empty path)
16:00:44 jaypipes dansmith: k, changed to +2 and recheck'd all the patches in that series up to the WIP ones.
16:01:10 dansmith jaypipes: thanks.. I figure they might need a rebase anyway, but wanted to make sure we resolved that -1 before I did
16:05:45 mriedem dansmith: is there a reason we can't hide that in the Instance object?
16:06:06 dansmith mriedem: hide what?
16:06:06 mriedem to generically use the cell mapping context when lazy-loading attrs?
16:06:32 dansmith mriedem: well, the lazy loading code can be run inside the cell too, which can't look up mappings
16:07:04 mriedem ok so it's done in the view because that's api which we know has an api db connection
16:07:24 dansmith right
16:13:14 mriedem dansmith: do you want to make a comment on https://review.openstack.org/#/c/405948/ about using the network_info for this instead of a new column?
16:13:35 dansmith sure
16:21:07 openstackgerrit Neil Jerram proposed openstack/nova: libvirt: avoid generating script with empty path https://review.openstack.org/411936
16:30:05 cdent rfolco: I'm back now, you get things working like you want?
16:38:32 rfolco cdent, not yet, was stuck on a call, then I had to restore my environment, so running tox -e for the first time takes forever
16:39:07 cdent rfolco: heh, yeah, that can be painful
16:39:19 rfolco cdent, I am adding sys.stderr.write as you suggested....
16:40:06 cdent Are you having issues with getting gabbi to behave like you want, or is that the handler code you're writing isn't doing what you expect?
16:41:31 rfolco cdent, both. I added trhe DELETE method to call set_inventories, just like PUT does. I created a delete_all in gabbits, but I am getting HTTP 500 response...
16:43:22 cdent when you get the 500 you'll get two different tracebacks: one from gabbi saying the test failed (this is the client side of the wsgi-intercept), and then a bit further up, the one from the wsgi app itself (the server side). Usually the latter is the one that will actually be useful and point out where things have actually gone wrong
16:45:28 rfolco cdent, I assume the gabbits should run in sequence so the delete works only if you create a resource and inventory first for example.
16:46:13 cdent rfolco: yes, that's right: each file is one sequence. even if you ask for just one test, the one's before it in the same file will be run, to make sure those prereqs are done
16:46:35 cdent If you want to push your code to gerrit as a wip I can look at it too?
16:47:18 rfolco cdent, let me do my homework, won't bother you much at this point. It's been a nice learning process.
16:47:54 cdent sure, but if you get to a point where it is no fun, feel free to ping me. Is no bother at all.
16:48:11 rfolco cdent, thanks for the support
16:57:12 openstackgerrit Stephen Finucane proposed openstack/nova: Stop handling differences in registerCloseCallback https://review.openstack.org/416297
17:05:08 mriedem sfinucan: see my reply in https://review.openstack.org/#/c/415707/ ?
17:06:08 sfinucan mriedem: I did not. Checking
17:06:56 mriedem sfinucan: basically i'd rather not subclass and duplicate the non-network-specific tests
17:07:15 sfinucan mriedem: gotcha
17:07:17 mriedem those quota tests are hittin the nova-net floating IP manager code,
17:07:32 sfinucan just those two, right?
17:07:37 mriedem yeah i think so,
17:08:06 mriedem so i think what you were suggesting was the base class handles the non-network quota stuff, and that runs with either nova-net or neutron config, and then sublcass that to just run the nova-net specific tests
17:08:15 mriedem but that also duplicates the network agnostic quota tests
17:08:38 mriedem it could be a separate test class altogether, idk, doesn't really seem worth the churn
17:08:46 sfinucan yeah, exactly
17:08:47 sfinucan but
17:08:57 sfinucan I didn't realise the other tests were actually network agnostic
17:09:39 sfinucan I thought they were executing different (but duplicated) code paths for some reason
17:10:29 sfinucan Lemme take a second look at it. Defo not worth the churn if what you say is correct
17:24:31 mriedem dansmith: while you're looking at newton-y things :) https://review.openstack.org/#/c/408727/
17:24:50 dansmith ack
17:24:55 openstackgerrit Chris Dent proposed openstack/nova: Add an "Adding a Handler" section to placement_dev https://review.openstack.org/416304
17:24:58 mriedem ^ is something i'd like to get in and backported to newton so we can release it as it's an upgrade issue
17:27:30 openstackgerrit Chris Dent proposed openstack/nova: [placement] Add an "Adding a Handler" section to placement_dev https://review.openstack.org/416304
17:38:02 sfinucan mriedem: Saw the error of my ways
17:48:19 openstackgerrit Claudiu Belu proposed openstack/nova: Add support for setting boot order in Hyper-V https://review.openstack.org/248029
18:00:30 openstackgerrit Diana Clarke proposed openstack/nova: api-ref: use the examples with paging links https://review.openstack.org/416312
18:17:01 stewie925 happy new year everyone
18:17:52 stewie925 hey I was trying to rebase a merge conflict - fixed the conflict but ran into a test issue - http://paste.openstack.org/show/593804/
18:21:27 jaypipes stewie925: source .tox/epy27/bin/activate && pip install -r test-requirements
18:21:54 stewie925 jaypipes: thank you so much
18:22:02 jaypipes stewie925: any time :)
18:22:40 jaypipes stewie925: sorry, that should be:
18:22:43 jaypipes stewie925: source .tox/py27/bin/activate && pip install -r test-requirements
18:22:53 jaypipes (no e in the tox env...)
18:23:05 stewie925 ahh hehe :)
18:26:56 stewie925 jaypipes: it erroed with "Could not open requirements file: [Errno 2] No such file or directory: 'test-requirements' "
18:27:29 jaypipes stewie925: you running this from the root of your nova development tree?

Earlier   Later