Open hjacobs opened 5 years ago
We're seeing this on k8 1.16.8, freshly installed (masters haven't been touched) following the readme instructions. I haven't been able to figure out the issue yet.
Experienced the same issue with EKS 1.16 on Fargate. It seems it is something to do with the CNI initialization.
As a workaround I added a sleeper init container to wait for 10 seconds
initContainers:
- name: wait-for-network
image: busybox:1.28
command: ['sh', '-c', "sleep 10"]
resources:
limits:
memory: 100Mi
requests:
cpu: 10m
memory: 100Mi
We see some "connection refused" errors after rolling master nodes. To be investigated.