support ARM and AMD for calico

[jayanthvn]

What type of PR is this?
enhancement

Which issue does this PR fix:
#1218

What does this PR do / Why do we need it:
Update calico-node and calico-typha to docker images to support ARM and AMD instances
Update autoscalar image - https://github.com/kubernetes-sigs/cluster-proportional-autoscaler/releases/tag/1.8.3

If an issue # is not available please add repro steps and logs from IPAMD/CNI showing the issue:

Testing done on this change:
Yes

AMD instance

kubectl describe node | grep Arch
 Architecture:               amd64

kgpsys | grep calico
calico-node-njjl2                                     1/1     Running   0          97m    192.168.7.155   ip-192-168-7-155.us-west-2.compute.internal   <none>           <none>
calico-typha-97c66d8b9-kdrdw                          1/1     Running   0          97m    192.168.7.155   ip-192-168-7-155.us-west-2.compute.internal   <none>           <none>
calico-typha-horizontal-autoscaler-6df548d5d5-5s9pp   1/1     Running   0          97m    192.168.29.9    ip-192-168-7-155.us-west-2.compute.internal   <none>           <none>

kubectl describe pod calico-node-njjl2 -n kube-system | grep Image
    Image:          calico/node:v3.15.1
    Image ID:       docker-pullable://calico/node@sha256:b386769a293d180cb6ee208c8594030128a0810b286a93ae897a231ef247afa8

kubectl describe pod calico-typha-97c66d8b9-kdrdw -n kube-system | grep Image
    Image:          calico/typha:v3.15.1
    Image ID:       docker-pullable://calico/typha@sha256:50830f75be50bb5a835c8705bce8745513374ce1cf1714af9b9321412f9b516f

kubectl describe pod calico-typha-horizontal-autoscaler-6df548d5d5-5s9pp -n kube-system | grep Image
    Image:         k8s.gcr.io/cpa/cluster-proportional-autoscaler:1.8.3
    Image ID:      docker-pullable://k8s.gcr.io/cpa/cluster-proportional-autoscaler@sha256:67640771ad9fc56f109d5b01e020f0c858e7c890bb0eb15ba0ebd325df3285e7

ARM instance

kubectl describe node | grep Arch
 Architecture:               arm64

kgpsys | grep calico
calico-node-26qkn                                     1/1     Running   0          6h9m   192.168.46.224   ip-192-168-46-224.us-west-2.compute.internal   <none>           <none>
calico-typha-6fb6797dc7-mpjtp                         1/1     Running   0          6h9m   192.168.46.224   ip-192-168-46-224.us-west-2.compute.internal   <none>           <none>
calico-typha-horizontal-autoscaler-6f6bd5b5df-ffhk5   1/1     Running   0          6h9m   192.168.42.169   ip-192-168-46-224.us-west-2.compute.internal   <none>           <none>

kubectl describe pod calico-node-26qkn -n kube-system | grep Image
    Image:          calico/node:v3.15.1
    Image ID:       docker-pullable://calico/node@sha256:b386769a293d180cb6ee208c8594030128a0810b286a93ae897a231ef247afa8

 kubectl describe pod calico-typha-6fb6797dc7-mpjtp -n kube-system | grep Image
    Image:          calico/typha:v3.15.1
    Image ID:       docker-pullable://calico/typha@sha256:50830f75be50bb5a835c8705bce8745513374ce1cf1714af9b9321412f9b516f

kubectl describe pod calico-typha-horizontal-autoscaler-6f6bd5b5df-ffhk5 -n kube-system | grep Image
    Image:         k8s.gcr.io/cpa/cluster-proportional-autoscaler:1.8.3
    Image ID:      docker-pullable://k8s.gcr.io/cpa/cluster-proportional-autoscaler@sha256:67640771ad9fc56f109d5b01e020f0c858e7c890bb0eb15ba0ebd325df3285e7

Automation added to e2e:

No
Will this break upgrades or downgrades. Has updating a running cluster been tested?:
No

Does this change require updates to the CNI daemonset config files to work?:

No

Does this PR introduce any user-facing change?:

Arm support for calico

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.

[jayanthvn]

/cc @caseydavenport can you please take a look?

[caseydavenport]

You will probably want to explicitly call out the registry here. e.g., docker.io/calico/node:v3.16.2

[jayanthvn]

Yes will do that

[gfvirga]

This is awesome! I spent an hour trying to figure why it didn't work for graviton, then I saw dockerhub had both arm64 and x86_64 architecture listed for the same tag and quay didn't decided to try. I have a cluster with both Archs and wanted the daemonset to run on all nodes

I got it working with helm upgrade --install --set calico.node.image="calico/node" --namespace kube-system aws-calico eks/aws-calico.

I am going to put a PR for the helm charts!

[rayterrill]

Just ran into this exact same issue. :(

[caseydavenport]

I haven't tried this version myself, but it should probably be fine.

[caseydavenport]

A couple minor comments from me but this LGTM.

[jayanthvn]

A couple minor comments from me but this LGTM.

Thanks for your time on reviewing the PR, I have made the change you suggested. Will merge the PR.