Wednesday, 2017-09-27

*** colettecello has joined #openstack-swg04:25
*** flaper87 has quit IRC04:29
*** stephenfin has quit IRC04:29
*** gothicmindfood has quit IRC04:29
*** stephenfin has joined #openstack-swg04:30
*** flaper87 has joined #openstack-swg04:37
*** jpich has joined #openstack-swg07:29
johnthetubaguyflaper87: I have attempted to clarify on the Glare front what would make me move to a +1 vote: https://review.openstack.org/#/c/47928514:20
flaper87johnthetubaguy: thanks a lot, I'll read and comment!14:21
flaper87:)14:21
johnthetubaguyflaper87: I would to hear if you think that makes any sense or not, its not all 100% clear in my head14:21
johnthetubaguywell, its proving hard to get what is in my head into words, as ever14:22
persiajohnthetubaguy: My memory of the conversation at the summit was that we wanted to keep Glance v2 API for a minimum of 5 years, starting from Queens release, and that everyone from Glare and Glance were willing to do that.  At the beginning of the discussion, Glare was hoping to introduce an images v3 earlier (perhaps with a mix of glance and glare features), but after the Nova implications and history were explained, the situation seemed calmer.14:49
persiaI'm not advocating you change your vote or position, but did want to make sure you understood that the issue was discussed, and the current Glance team supports your view (and Glare seemed to be willing to consider it).14:50
johnthetubaguypersia: thanks for the context. I am more on the side of forever, rather than for 5 years.14:51
johnthetubaguyto be clear, that is all about supporting the existing REST requests that succeed today14:51
johnthetubaguythere can be new fancy requests that do better things, I have no problem with all that14:51
persiaSo were most of the Glance folk :)  The main reason for retreating to 5 years was the idea that we couldn't know the future, and there are some inefficiencies that may be hard to work around with v2, plus the existence of the experimental v3 some years back.14:52
persiaUnderstood.  I believe rosmaita will support your position as long as he remains PTL, and I suspect his successor to do the same unless there is a *really* good reason that is plain also to Nova folk.14:53
johnthetubaguyI don't think inefficiencies would ever be bad enough to make all our users re-write their scripts14:54
johnthetubaguypersia: good to hear most folks are supporting the direction of the stable public API though14:56
persiaI don't like to make multi-year future statements, but yes, I hope there is never a good reason :)14:57
johnthetubaguyI think sdague's blog post is the best expression of where I am coming from: https://dague.net/2015/06/05/the-nova-api-in-kilo-and-beyond-2/14:58
persiaRight.  My limited understanding is that there is no path for Olivia with the current glance v2 API.  I would be delighted to be mistaken about that.15:00
johnthetubaguypersia: you can add microversion to an existing API, allowing you to expose the new features earlier15:08
persiaPerhaps because of the V1/V2 pain, I had a feeling the glance team was nervous about that, but if it is true, then I suspect there will never be a glance v3 API15:09
johnthetubaguythat was mostly why I brought up the image lifecycle problem, and how that gets fixed15:09
johnthetubaguyNova added micro-versions to its existing API to avoid the need for a big v2 jump15:10
persiaRight.  Image lifecycle was a big priority for glance this cycle, but that was one of the glance sessions I missed.15:10
johnthetubaguyI couldn't make any of them sadly, too many other places to be :(15:14
*** jpich has quit IRC16:16
*** alexismonville has left #openstack-swg19:29
*** colettecello has quit IRC19:47
*** gothicmindfood has joined #openstack-swg19:47

Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!