thread 31: kubernetes test for every merge thread 62: plow proud plug-in not provider specific 62: i think there's also something that we can all think of going forward as how to scale how toScale the lkg testing as we as we all adopted. i think one of the ways we could do that is by sharing kubernetes builds. i'm sure there will be some questions i have a go ahead michelle okay i have a question does [ play ]
thread 68: kubernetes support for all versions 68: it's just going to be a lower volume on those branches so i was actually going to suggest i believe andrew has a separate document that talks about branching requirements for cloud providers. i [ play ]
thread 100: How can we use kk to run tests in UH? 100: cube up is not requiring cube up to be implemented by everyone but at the same time we get a well-defined list of things like what you need to give back whether it's a cube config file ingresses whatever access tokens etc. The other part to this that i think is worth talking about is everything we've said is sort of going from pre [ play ]
#what #whateveraccesstokens #joe
thread 104: Angular2: Should we automatically roll back on one of these? 104: If you have a last known good wouldn't we expect that to be used to set up a cluster that has the same capabilities that it has today only using the out of tree gcp provider. It would still run all those storage tests that you already have today it would certainly run them and in fact i i've talked to tim hawkin and a few others it theoretically could run them as priests submit. [ play ]
thread 6: i'm going to be opening up for screen sharing thread 17: kubernetes: how to run tests that aren't about cloud providers thread 63: sharing kubernetes builds 63: computed storage of all these different builds of kubernetes there will be other ways to that we can all probably share the load in one way or another going forward so yeah i think that was about was about it so if you all had questions i think we can go ahead [ play ]
#allthesedifferentbuilds
thread 26: Kubernetes lkg is always present but not in the background thread 79: how to build a client library that isn't changing? 79: i get that the one interesting alternate view on this is that in addition to the normal libraries we think of like client go the other piece that you may be pulling or not maybe you will be pulling in in the ccm include sig cloud provider or even let's say storage provided controllers. i i think that makes sense [ play ]
thread 33: kubernetes post submit and periodic commit thread 95: How do you test all the things that need to be tested on kubernetes 95: Lkg is just this thing that's helping us keep things recent but it's not actually solving the problem of like all the things that we need to test so we're gonna have to get all that figured out. We're going to start with a subset which is e to e and conformance and then work our way out. [ play ]
thread 102: What is the difference between pre-submit and post submit? 102: The question becomes how do we consume breakages like is it just shows up as a proud as in some sort of proud job or and when things break at what point are we willing to say we should revert the pr that caused this right. i think there are two opposite views in my mind and i think t are both valid. [ play ]
#proudjob #whatpoint #thisonecloudprovider
thread 108: gcp cloud provider last known good 108: i understand why the last known good calculation that you were trying to talk about in in the presentation can't be done on pre-submit because that may require rebuilding and working but if we already have a shaw or an image pull stack i'm not clear on why you would need to rebuild the gcp [ play ]
#thepresentation #thegcpcloudprovider #what
thread 14: run cloud provider specific tests and get signal back thread 55: lkg tracking multiple cloud providers 55: The lkg version will be independent of each other something else that'll have to be taken into account. There's a couple top provider specific components that everyone should be aware of. It's pretty it's preoccupied with building and testing each cloud provider so it's going to be very specific to each one. [ play ]
thread 29: kubernetes downstream test against cloud providers thread 98: how do i get the test running bit to work? 98: i think we already know how to start everything up so yeah we'll try and get get that circulated as part of the design dock cool thank you i'd actually like to keep it seer and slurry's hand i think one of the things that it would be good to do would be to try to have a general interface where we say i cloud provider can put in some minimal thing into the kk which can then link into [ play ]
thread 36: i can't start screen sharing thread 45: lkvg version tracking thread 120: cloud provider extraction work matt 120: Csi migration for gce where we are starting in fact to figure out which which tests are going to need to be moved and in particular issues where like some tests that depend on gce maybe i actually don't we could use like a a host path driver. i don't want to commit them to to necessarily solving the problem. [ play ]