| openstackgerrit | Merged openstack/defcore: Added keystone scoring for Mitaka release cycle. https://review.openstack.org/297847 | 03:45 |
|---|---|---|
| openstackgerrit | Merged openstack/defcore: fix some typos/punctuation in CoreCriteria.rst https://review.openstack.org/307153 | 03:55 |
| openstackgerrit | Merged openstack/defcore: Leave capabilities advisory until tests updated https://review.openstack.org/305269 | 04:01 |
| *** markvoelker has quit IRC | 04:34 | |
| *** markvoelker has joined #openstack-defcore | 04:35 | |
| *** markvoelker has quit IRC | 04:40 | |
| openstackgerrit | Chris Hoge proposed openstack/defcore: Added 2016A Swift Scoring Matrix https://review.openstack.org/290592 | 04:52 |
| openstackgerrit | Chris Hoge proposed openstack/defcore: Added 2016A Swift Scoring Matrix https://review.openstack.org/290592 | 05:01 |
| openstackgerrit | Chris Hoge proposed openstack/defcore: Added 2016A Swift Scoring Matrix https://review.openstack.org/290592 | 05:02 |
| *** pilgrimstack has joined #openstack-defcore | 06:42 | |
| *** pilgrimstack has quit IRC | 07:11 | |
| *** pilgrimstack has joined #openstack-defcore | 08:06 | |
| *** markvoelker has joined #openstack-defcore | 12:18 | |
| *** cjvolzka has joined #openstack-defcore | 12:58 | |
| *** edmondsw has joined #openstack-defcore | 13:38 | |
| *** amit213 has quit IRC | 15:06 | |
| *** amit213 has joined #openstack-defcore | 15:08 | |
| openstackgerrit | Chris Hoge proposed openstack/defcore: Added cinder scoring for Mitaka release cycle https://review.openstack.org/290271 | 15:54 |
| *** lcazares has joined #openstack-defcore | 15:56 | |
| *** lcazares has quit IRC | 15:57 | |
| *** lcazares has joined #openstack-defcore | 15:59 | |
| *** galstrom_zzz is now known as galstrom | 16:29 | |
| docaedo | man, sorry I missed a really interesting 20 minute section of the meeting today :( | 17:08 |
| hogepodge | To be clear, if a fixture is an interop capability, I have no problem with failing a test because that fixture fails. Not saying there's not room for improvement on the test, but free passes should be given for technicalities. | 17:14 |
| hogepodge | If a fixture is not an interop capability there's a strong case to not require it until the problem is fixed. | 17:15 |
| markvoelker | hogepodge: and, in a perfect world, if a test has a fixture that's another required capability, that required capability would also have a test, right? | 17:19 |
| markvoelker | Which would probably fail in most of the cases where the fixture failed.... | 17:19 |
| docaedo | The part I don't understand is that what matters is "can users all make use of a certain ability" and if you can make a volume, but can't get rid of it, that's only half the feature | 17:21 |
| docaedo | "identify a need for an unmount test" is the part I would have been disagreeing with - yes, there's a need for that, but it should all be part of "do volumes work here?" | 17:22 |
| markvoelker | docaedo: typing while in a meeting here, so you'll have to pardon slow replies. Maybe some background is useful: | 17:24 |
| markvoelker | docaedo: so, the tricky bit is how we define a "Capability" and how we test that Capability. You can argue one Capability = 1 API, or you could argue that multiple API's (e.g. everything in a CRUD) = 1 Capability. | 17:25 |
| hogepodge | markvoelker one would hope so | 17:25 |
| *** lcazares has quit IRC | 17:25 | |
| markvoelker | docaedo: also important to consider multiple ways to do things | 17:25 |
| markvoelker | docaedo: for example, there are some tests that create images with glance v1 as part of a fixture | 17:26 |
| markvoelker | If we require v2 but don't require v1... | 17:26 |
| markvoelker | etc etc | 17:26 |
| hogepodge | markvoelker if that's not the case, my preferred course of action is to count test as fail and write new test to cover gap, and not flag on technicality | 17:26 |
| *** lcazares has joined #openstack-defcore | 17:26 | |
| markvoelker | Or if a fixture uses the import task vs image create | 17:27 |
| docaedo | I can appreciate how it gets tricky - I think from my perspective the value of defcore/refstack/interop though is that users have the same experience across openstack clouds | 17:27 |
| docaedo | and that's tough because openstack allows a dozen competing ways to solve the same problem, and operators want flexibility to do whatever they feel like doing (but they still want to be certified anyway) | 17:28 |
| *** MarkAtwood has joined #openstack-defcore | 17:29 | |
| markvoelker | docaedo: right. And so in some cases we might get into situations where, surprisingly, the market hasn't clearly picked a winner among the various ways to do some part of a CRUD operation, sadly. =/ | 17:29 |
| docaedo | (I'm not driving towards a great point with a perfect answer here, FYI :) Just trying to understand more of the background and politics around this stuff) | 17:29 |
| markvoelker | Oh, you don't have a perfect answer? Well, you'll fit right in around here then. =) | 17:30 |
| docaedo | I've finally found my home! | 17:30 |
| * markvoelker has to tune back into his meeting for a bit | 17:31 | |
| hogepodge | markvoelker import task is deprecated and admin only. It is not interop. If you watch the Vancouver summit video about the task API, the speaker acknowledges the task api is not discoverable or interoperable | 18:02 |
| markvoelker | hogepodge: Sure, just an example. Maybe image-list is a better one. | 18:03 |
| hogepodge | We flagged in that instance. One year later the underlying problem is not fixed. | 18:05 |
| hogepodge | We tried to set policies to encourage upstreaming and fixing by vendors, but those policies seem to have not provided the proper incentives to do that. | 18:07 |
| markvoelker | I'd argue that there's been progress toward it though. Lots of discussion about dropping the nova proxy API's and making that policy, glance moving toward depracting v1, etc. | 18:08 |
| hogepodge | The immediate issue was a v2 proxy, which is desired but still a wish list item | 18:08 |
| hogepodge | Maybe requiring direct image will fix that, but those who depend on task api will probably object and the unified APIs is also wip, making it a long term interop candidate | 18:09 |
| hogepodge | So, as of today, one of the most important aspects of cloud has no interoperable interface. | 18:10 |
| *** lcazares is now known as Guest74905 | 18:11 | |
| *** luzC is now known as lcazares | 18:11 | |
| markvoelker | Yup. And there is no way to fix that overnight. | 18:11 |
| hogepodge | Networking will be another can of worms. Some clouds don't want to give you any control over network, and we have no good way to guarantee network connectivity in an interop way | 18:12 |
| markvoelker | All fine candidates for our forthcoming report on major interop issues that OpenStack has today. =) | 18:14 |
| *** MarkAtwood has quit IRC | 18:15 | |
| *** lcazares has quit IRC | 18:15 | |
| *** luzC has joined #openstack-defcore | 18:16 | |
| *** luzC- has joined #openstack-defcore | 18:19 | |
| *** luzC- has quit IRC | 18:20 | |
| *** luzC- has joined #openstack-defcore | 18:24 | |
| *** luzC has quit IRC | 18:25 | |
| *** luzC- is now known as luzC | 18:25 | |
| *** lcazares has joined #openstack-defcore | 18:28 | |
| *** galstrom has quit IRC | 18:35 | |
| *** galstrom_zzz has joined #openstack-defcore | 18:35 | |
| *** galstrom_zzz is now known as galstrom | 18:35 | |
| *** woodster_ has joined #openstack-defcore | 18:39 | |
| *** galstrom has quit IRC | 19:00 | |
| *** galstrom has joined #openstack-defcore | 19:00 | |
| openstackgerrit | Merged openstack/defcore: Added 2016A Swift Scoring Matrix https://review.openstack.org/290592 | 20:32 |
| *** MarkAtwood has joined #openstack-defcore | 21:00 | |
| openstackgerrit | Chris Hoge proposed openstack/defcore: Added cinder scoring for Mitaka release cycle https://review.openstack.org/290271 | 21:01 |
| hogepodge | markvoelker: eglute: ^^ | 21:01 |
| *** luzC has quit IRC | 21:16 | |
| *** luzC has joined #openstack-defcore | 21:21 | |
| *** edmondsw has quit IRC | 22:24 | |
| *** galstrom is now known as galstrom_zzz | 22:36 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!