Calico Networking Fails To Initialize on New or Upgraded AWS-based FedoraCoreOS Next/Testing Typhoon Clusters

[fifofonix]

Description

Calico networking pods fail to initialize on Typhoon clusters in AWS running FedoraCoreOS Next/Testing.

Steps to Reproduce

Provide clear steps to reproduce the bug.

Provisioning clusters via tutorial on website for AWS/FedoraCoreOS except:

• Using a private hosted zone and networking = "calico"

Then:

• Follow tutorial with `os_stream = "stable" (default) shows cluster members in a Ready status. Success.
• Follow tutorial with `os_stream = "testing", cluster members do not get to a Ready status. Calico pods fail init.

Logs captured from the failing cluster:

kubectl logs pod/calico-node-xxxx -c install-cni -n kube-system

2024-12-09 23:02:50.331 [INFO][1] cni-installer/<nil> <nil>: Running as a Kubernetes pod
2024-12-09 23:02:50.346 [INFO][1] cni-installer/<nil> <nil>: File is already up to date, skipping file="/host/opt/cni/bin/bandwidth"
2024-12-09 23:02:50.347 [INFO][1] cni-installer/<nil> <nil>: Installed /host/opt/cni/bin/bandwidth
2024-12-09 23:02:50.481 [INFO][1] cni-installer/<nil> <nil>: File is already up to date, skipping file="/host/opt/cni/bin/calico"
2024-12-09 23:02:50.481 [INFO][1] cni-installer/<nil> <nil>: Installed /host/opt/cni/bin/calico
2024-12-09 23:02:50.624 [INFO][1] cni-installer/<nil> <nil>: File is already up to date, skipping file="/host/opt/cni/bin/calico-ipam"
2024-12-09 23:02:50.624 [INFO][1] cni-installer/<nil> <nil>: Installed /host/opt/cni/bin/calico-ipam
2024-12-09 23:02:50.629 [INFO][1] cni-installer/<nil> <nil>: File is already up to date, skipping file="/host/opt/cni/bin/flannel"
2024-12-09 23:02:50.630 [INFO][1] cni-installer/<nil> <nil>: Installed /host/opt/cni/bin/flannel
2024-12-09 23:02:50.637 [INFO][1] cni-installer/<nil> <nil>: File is already up to date, skipping file="/host/opt/cni/bin/host-local"
2024-12-09 23:02:50.637 [INFO][1] cni-installer/<nil> <nil>: Installed /host/opt/cni/bin/host-local
2024-12-09 23:02:50.648 [INFO][1] cni-installer/<nil> <nil>: File is already up to date, skipping file="/host/opt/cni/bin/loopback"
2024-12-09 23:02:50.648 [INFO][1] cni-installer/<nil> <nil>: Installed /host/opt/cni/bin/loopback
2024-12-09 23:02:50.656 [INFO][1] cni-installer/<nil> <nil>: File is already up to date, skipping file="/host/opt/cni/bin/portmap"
2024-12-09 23:02:50.656 [INFO][1] cni-installer/<nil> <nil>: Installed /host/opt/cni/bin/portmap
2024-12-09 23:02:50.664 [INFO][1] cni-installer/<nil> <nil>: File is already up to date, skipping file="/host/opt/cni/bin/tuning"
2024-12-09 23:02:50.664 [INFO][1] cni-installer/<nil> <nil>: Installed /host/opt/cni/bin/tuning
2024-12-09 23:02:50.664 [INFO][1] cni-installer/<nil> <nil>: Wrote Calico CNI binaries to /host/opt/cni/bin

2024-12-09 23:02:50.779 [INFO][1] cni-installer/<nil> <nil>: CNI plugin version: v3.27.3
2024-12-09 23:02:50.779 [INFO][1] cni-installer/<nil> <nil>: /host/secondary-bin-dir is not writeable, skipping
2024-12-09 23:02:50.780 [WARNING][1] cni-installer/<nil> <nil>: Neither --kubeconfig nor --master was specified.  Using the inClusterConfig.  This might not work.
2024-12-09 23:03:20.781 [ERROR][1] cni-installer/<nil> <nil>: Unable to create token for CNI kubeconfig error=Post "https://10.3.0.1:443/api/v1/namespaces/kube-system/serviceaccounts/calico-cni-plugin/token": dial tcp 10.3.0.1:443: i/o timeout
2024-12-09 23:03:20.781 [FATAL][1] cni-installer/<nil> <nil>: Unable to create token for CNI kubeconfig error=Post "https://10.3.0.1:443/api/v1/namespaces/kube-system/serviceaccounts/calico-cni-plugin/token": dial tcp 10.3.0.1:443: i/o timeout

Additional information:

Testing 41.20241122.2.0 (kernel 6.11.8-300.fc41.x86_64) is the first testing version to experience this issue. Current next 41.20241122.1.0 also experiences the issue, but I think the prior version does too but that would require further tests to confirm.

This issue first occurred in our much customized typhoon clusters in non-prod which are a fork of typhoon v1.30 so I don't think this issue is related to the most recent version of k8s and/or typhoon. We have recreated these clusters using "ciluim" networking instead for the time being and this has addressed issues.

Expected behavior

Success

Environment

• Platform: aws
• OS: fedora-coreos
• Release: v1.31.2
• Terraform: v1.5.7
• Plugins:

- Installing hashicorp/local v2.5.2...
- Installed hashicorp/local v2.5.2 (signed by HashiCorp)
- Installing hashicorp/null v3.2.3...
- Installed hashicorp/null v3.2.3 (signed by HashiCorp)
- Installing hashicorp/random v3.6.3...
- Installed hashicorp/random v3.6.3 (signed by HashiCorp)
- Installing hashicorp/tls v4.0.6...
- Installed hashicorp/tls v4.0.6 (signed by HashiCorp)
- Installing poseidon/ct v0.13.0...
- Installed poseidon/ct v0.13.0 (self-signed, key ID 8F515AD1602065C8)
- Installing hashicorp/aws v4.61.0...
- Installed hashicorp/aws v4.61.0 (signed by HashiCorp)

Possible Solution

Link to a PR or description.

[fifofonix]

This sounds similar: projectcalico/calico#8368