conformance: support non-amd64 architectures

[howardjohn]

Similar to #627, we should publish all images used in conformance tests with multi-arch support.

One thing that confused me is @mikemorris seemed to suggest we picked the image we use because it supports arm64: kubernetes/kubernetes#110605 (comment). However, I do not see the same:

$ crane manifest k8s.gcr.io/ingressconformance/echoserver:v0.0.1
{
   "schemaVersion": 2,
   "mediaType": "application/vnd.docker.distribution.manifest.v2+json",
   "config": {
      "mediaType": "application/vnd.docker.container.image.v1+json",
      "size": 2026,
      "digest": "sha256:3cb9c24fedce680ad075cdc5d3d4037e306217fa1824344a54a0bd1d65a98654"
   },
   "layers": [
      {
         "mediaType": "application/vnd.docker.image.rootfs.diff.tar.gzip",
         "size": 639433,
         "digest": "sha256:5c7fe08dec514a5105d43dfcae1a5a45be002bcd3251dd178ced702aecfb7531"
      },
      {
         "mediaType": "application/vnd.docker.image.rootfs.diff.tar.gzip",
         "size": 2317481,
         "digest": "sha256:d6a250117705cac0781f5218e3753ba42a048ebbc49150b39e9a42dce0b189ef"
      }
   ]
}

[howardjohn]

I think the source is https://github.com/kubernetes-sigs/ingress-controller-conformance/tree/master/images/echoserver - It should be easy to change and I am happy to do it, just not sure how to actually get it published

[howardjohn]

/assign @howardjohn

I'll need help getting it published officially though

[mikemorris]

Hmm, interesting that we had found this to work in practice previously without explicitly building arm64 images - not quite sure how it was working without digging deeper, but hashicorp/learn-consul-kubernetes#65 was the behavior we found where switching to this image was working on an M1 chip.

[howardjohn]

possibly using emulation?

…

On Mon, Aug 22, 2022, 11:50 AM Mike Morris ***@***.***> wrote: Hmm, interesting that we had found this to work in practice previosuly without explicitly building arm64 images - not quite sure *how* it was working without digging deeper, but hashicorp/learn-consul-kubernetes#65 <hashicorp/learn-consul-kubernetes#65> was the behavior we found where switching to this image was working on an M1 chip. — Reply to this email directly, view it on GitHub <#1295 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAEYGXJKQQWEC6UNMNBK4L3V2PDXNANCNFSM55TBW2IA> . You are receiving this because you were assigned.Message ID: ***@***.***>