Michael Bell
Michael Bell
Given the above, is this - A bug in istio injector not merging the two correctly (there's no conflict per se) - A bug in the docs, and the only...
I'd be happy to try to submit clearer docs, but it's not obvious to me the current behavior is as designed.
1. We have a chart to deploy our applications. 2. It always specifies proxy cpu/memory requests and limits. Similar to ``` sidecar.istio.io/proxyMemory: {{ .memory | quote }} sidecar.istio.io/proxyMemoryLimit: {{ .memory...
Part of the confusion in this ticket @howardjohn is we have two completely different templates. We have the go-template injection by the injection controller and we have the helm template...
To further clarify. Here's fhe first part of our template pod template spec. I've edited it down ``` template: metadata: annotations: app.kubernetes.io/version: {{ required ".Values.image.tag is required " .Values.image.tag |...
[Archive.zip](https://github.com/user-attachments/files/17997106/Archive.zip) Yes. Attached are rendered deployments and the final injected pods. Note memory and cpu use the "default" with prestop, and use the annotations without
That makes sense! But why does it only kick in when the skeleton option exists? I'm guessing because the yaml is wholly absen in that case So without our "template"...
Those are all good ways. But seem inflexible. I'm not complaining per se - the design issues is fundamental to order of operations of admissions controllers, but very much affects...