Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2017-02-14
09:37:10 johnthetubaguy alex_xu: ideally it would have got into some distro somewhere, so it could be tested
09:37:37 alex_xu johnthetubaguy: got it, thanks
09:37:40 johnthetubaguy alex_xu: we might talk about that again at the PTG, we need to get some way of testing that (I am currently wondering if ubuntu cloud archive would get us closer)
09:38:06 johnthetubaguy fedora bleeding edge thingy whatsit was suggested, but never seems to have happened
09:38:18 johnthetubaguy alex_xu: oh, did you see I created some updated policy things: https://review.openstack.org/#/c/433010/
09:38:33 johnthetubaguy alex_xu: there are three specs instead of one spec now
09:38:49 johnthetubaguy ah, I see you commented on one, so that is probably a yes
09:38:58 alex_xu johnthetubaguy: yea, I just read them :)
09:39:36 alex_xu will try to think more
09:40:39 openstackgerrit huangtianhua proposed openstack/nova master: Don't delete neutron port when attach failed https://review.openstack.org/403568
09:47:22 johnthetubaguy alex_xu: cool, that middle one really turns things an interesting direction, got some great feedback from some keystone folks that made me come up with that idea
09:48:36 johnthetubaguy dstanak's comments I think it was
09:49:16 alex_xu johnthetubaguy: yes, custom scope would create interoperaibility issues is interesting
09:49:52 alex_xu johnthetubaguy: may i ask an example for that interoperability issue, try to understand that clearly
09:51:14 openstackgerrit OpenStack Proposal Bot proposed openstack/nova master: Updated from global requirements https://review.openstack.org/432001
09:51:26 johnthetubaguy alex_xu: there are two users both creating instances, they share project wide monitoring system to track their usage, and check they have the correct ports closed
09:51:46 johnthetubaguy alex_xu: they move to a different OpenStack cloud and it breaks, because each user can't see each others servers
09:52:42 johnthetubaguy we say servers are owned by a project, keypairs by a user, if that changes it makes the API quite different
09:55:28 alex_xu johnthetubaguy: thanks, I see now
09:59:05 alex_xu johnthetubaguy: dose it mean that don't break the user which is without special role?
10:00:16 alex_xu as user with different role, it still can be configured to have different visibility, some role have global visibility, the same role move to another cloud, may lose that visibility.
10:00:44 johnthetubaguy alex_xu: sorry, not sure what you mean
10:00:56 johnthetubaguy alex_xu: on different clouds, there may be different roles, for sure
10:01:23 johnthetubaguy alex_xu: my hope is most of those differences we allow can be exposed via the capabilities API
10:07:05 alex_xu johnthetubaguy: why we didn't use capabilities API to resolve the interoperability issues you mentioned above
10:16:24 andymccr mriedem_afk i added a comment to the PR incase we're not all on at the same time again - appreciate the feedback and everybody who has looked at the endpoint issue!
10:28:36 johnthetubaguy alex_xu: we can't report every single change in the capabilities API, this is just limiting the allowed changes to a reasonable set of things.
10:35:27 bauzas \o
10:37:10 openstackgerrit Balazs Gibizer proposed openstack/nova master: api-ref: note that boot ignores bdm:device_name https://review.openstack.org/433575
10:46:22 openstackgerrit Artom Lifshitz proposed openstack/nova master: Run api sample tests against 2.latest https://review.openstack.org/430352
10:46:23 openstackgerrit Artom Lifshitz proposed openstack/nova master: Stub network and volume api in api sample tests https://review.openstack.org/433584
10:46:23 openstackgerrit Artom Lifshitz proposed openstack/nova master: api sample tests: handle response missing an item https://review.openstack.org/431738
10:46:24 openstackgerrit Artom Lifshitz proposed openstack/nova master: Add skip_latest_microversion decorator https://review.openstack.org/433585
10:48:06 openstackgerrit Artom Lifshitz proposed openstack/nova master: Add skip_latest_microversion decorator https://review.openstack.org/433585
10:48:06 openstackgerrit Artom Lifshitz proposed openstack/nova master: Run api sample tests against 2.latest https://review.openstack.org/430352
10:48:07 openstackgerrit Artom Lifshitz proposed openstack/nova master: Stub network and volume api in api sample tests https://review.openstack.org/431723
10:48:07 openstackgerrit Artom Lifshitz proposed openstack/nova master: api sample tests: handle response missing an item https://review.openstack.org/431738
10:50:59 openstackgerrit Artom Lifshitz proposed openstack/nova master: Add skip_latest_microversion decorator https://review.openstack.org/433585
10:50:59 openstackgerrit Artom Lifshitz proposed openstack/nova master: Run api sample tests against 2.latest https://review.openstack.org/430352
10:51:00 openstackgerrit Artom Lifshitz proposed openstack/nova master: Stub network and volume api in api sample tests https://review.openstack.org/431723
10:51:00 openstackgerrit Artom Lifshitz proposed openstack/nova master: api sample tests: handle response missing an item https://review.openstack.org/431738
11:01:36 openstackgerrit Artom Lifshitz proposed openstack/nova master: Add skip_latest_microversion decorator https://review.openstack.org/433585
11:01:36 openstackgerrit Artom Lifshitz proposed openstack/nova master: Run api sample tests against 2.latest https://review.openstack.org/430352
11:01:37 openstackgerrit Artom Lifshitz proposed openstack/nova master: Fakes and stubs for 2.latest api sample tests https://review.openstack.org/431723
11:01:37 openstackgerrit Artom Lifshitz proposed openstack/nova master: api sample tests: handle response missing an item https://review.openstack.org/431738
11:10:46 bauzas johnthetubaguy: since I was a bit off last week, any urgent thing to review ?
11:15:25 johnthetubaguy bauzas: backports that are pending on ocata/stable?
11:15:52 bauzas johnthetubaguy: okay, just looking
11:17:20 johnthetubaguy snikitin: bit worried about the upgrade situation, given some ideas on how to use the service version mechanism to make that cleaner
11:17:24 bauzas johnthetubaguy: since we haven't cut a RC2 last week, I guess we're planning it for Thurs ?
11:17:35 johnthetubaguy bauzas: thats my assumption
11:17:54 johnthetubaguy bauzas: unless something big and nasty comes along
11:17:59 bauzas ok
11:18:08 bauzas okay, so looking at bugs then
11:18:14 bauzas and the stable branch
11:18:52 snikitin johnthetubaguy: thanks, I'll look
11:18:57 johnthetubaguy snikitin: I was also a bit unsure how the object works OK for the older microversion, that seemed more like chance? I must be missing something around that
11:26:04 snikitin johnthetubaguy: The older microversions use get_diagnostics() method. This method always returns a dicts. and the newest microversions use get_instance_diagnostics() methods. which returns an object. Or did you meant something else?
11:26:27 snikitin johnthetubaguy: The older microversions use get_diagnostics() method. This method always returns a dicts. and the newest microversions use get_instance_diagnostics() method. which returns an object. Or did you meant something else?
11:27:09 johnthetubaguy snikitin: doh, I totally misread that, so this is the first time get_instance_diagnostics is used?
11:27:18 snikitin yes
11:27:50 johnthetubaguy snikitin: OK, so thats probably easier, I would just bump the RPC API version, and error out if we are not above that version when calling get_instance_diagnostics
11:28:27 johnthetubaguy i.e. it must always return an object for the new method, or raise an exception
11:29:11 snikitin johnthetubaguy, ok. and what time of exception it should be? forbidden?
11:30:24 snikitin type*
11:33:56 johnthetubaguy snikitin: I can't remember what we do now, I will have a look
11:36:00 snikitin johnthetubaguy: omg. The version of rpcapi already bumped o_O https://review.openstack.org/#/c/333988/4/nova/compute/rpcapi.py
11:37:25 johnthetubaguy snikitin: the problem is, its changing again, so it needs to be done again
11:37:38 johnthetubaguy snikitin: code needs to know it is getting an object returned
11:37:56 johnthetubaguy snikitin: that bump was just for the getting an object into the method, I think
11:38:06 snikitin johnthetubaguy: ok! will do. thanks for review!
11:38:45 openstackgerrit Stephen Finucane proposed openstack/nova-specs master: Add 'move-nova-cmds-to-click' spec https://review.openstack.org/433603
11:39:57 johnthetubaguy snikitin: looks like we went for badrequest in the past: https://github.com/openstack/nova/blob/master/nova/api/openstack/compute/servers.py#L587
11:40:54 snikitin johnthetubaguy: Ok. will do the same
11:41:00 johnthetubaguy snikitin: thanks
11:57:44 openstackgerrit Stephen Finucane proposed openstack/nova master: hacking: Reword in the imperative https://review.openstack.org/363624
11:58:58 openstackgerrit John Garbutt proposed openstack/nova-specs master: Add polcy-docs spec https://review.openstack.org/433010
12:14:22 sdague TFW you make a change that should break cinder, and all the tests pass ... o_O
12:23:40 cdent :(
12:35:17 openstackgerrit John Garbutt proposed openstack/nova-specs master: Add policy-project-checks spec https://review.openstack.org/433037
12:40:35 openstackgerrit Artom Lifshitz proposed openstack/nova master: Run api sample tests against 2.latest https://review.openstack.org/430352
12:40:36 openstackgerrit Artom Lifshitz proposed openstack/nova master: api sample tests: handle response missing an item https://review.openstack.org/431738
12:40:36 openstackgerrit Artom Lifshitz proposed openstack/nova master: Add skip_latest_microversion decorator https://review.openstack.org/433585
12:40:37 openstackgerrit Artom Lifshitz proposed openstack/nova master: Fixtures for 2.latest api sample tests https://review.openstack.org/431723
12:43:11 openstackgerrit John Garbutt proposed openstack/nova-specs master: Add policy-project-checks spec https://review.openstack.org/433037
12:43:11 openstackgerrit John Garbutt proposed openstack/nova-specs master: Add polcy-docs spec https://review.openstack.org/433010
12:43:54 openstackgerrit John Garbutt proposed openstack/nova-specs master: Add policy-remove-scope-checks spec https://review.openstack.org/433037
12:49:36 openstackgerrit Rodolfo Alonso Hernandez proposed openstack/nova master: Optimize time during PCI alias parsing process https://review.openstack.org/427145
13:01:47 sdague cdent: yeh, well I sent an email to the list, will try to figure it out there
13:32:47 alex_xu bauzas: cdent why we need another message for api response than the exception one https://review.openstack.org/#/c/428612/3/nova/api/openstack/placement/handlers/resource_class.py@90?
13:33:37 alex_xu ensure whether it is on purpose
13:33:39 cdent alex_xu: just to be nice and explicit?
13:34:30 cdent basically the idea is that we want the http api error messages to be fairly detailed and informative but the generic/original exception doesn't need to be
13:34:42 alex_xu cdent: yeah, we need another one if the exception just talk about common thing or implement thing
13:35:38 alex_xu cdent: emm...probably for this case https://review.openstack.org/#/c/428612/3/nova/exception.py, they are very close i guess
13:36:22 cdent since we're going to reraise anyway (to make sure we are using the webob exception) having another message is fine, I would think?
13:42:08 bauzas cdent: well, I'm not really sure the detailed message really helps here
13:42:20 bauzas cdent: it says "sorry, please retry"
13:42:29 bauzas well, okay, but why ?
13:42:44 cdent bauzas: did you see my earlier -1 comment?

Earlier   Later