John Howard

Yep, exactly. There is probably a few approaches: * Don't use limit ranger (understandably unacceptable) * Add the resources into the `istio-proxy` container spec * Instead of in-lining the config...

> This is injected as permanent template in the injector CM I assume from what I read? That seems like that's not terribly flexible since you'd need to have 1...

Host checks are almost exclusively used in ingress so it kind of makes sense as written. Is there some context behind the motivation here?

So the root cause is a difference in how Helm directly merges vs how we are merging with the 'defaults' stuff. I think basically caused by this line: https://github.com/helm/helm/blob/847369c184d93fc4d36e9ec86a388b60331ab37a/pkg/chartutil/coalesce.go#L225. We...

> can you elaborate please? We do merging in our chart with `mergeOverwrite`. It behaves differently from how helm internally merges

Even if its a great idea its probably not one we will make from a backwards compatibility perspective. We do know a baseline expectation of usage, so it seems wrong...

I think that PR was just retaining the existing ordering which was probably accidental

Yangmin is no longer working on the project so I wouldn't expect a response On Sun, Jul 16, 2023 at 11:02 AM amar1104 ***@***.***> wrote: > Hi @yangminzhu , any...

Every day we see requests to add 1 more small hack to address 1 more small use case. In isolation each one is fine - if we accept 1 a...

Related: #46899 On Thu, Oct 5, 2023 at 7:35 PM Lin Sun ***@***.***> wrote: > I agree, we need to look at the requests holistically. Is someone willing > to...