If you find that your calico nodes are crashing on your DKP 1.X clusters, you may be hitting an issue described in detail here:
https://github.com/projectcalico/calico/issues/4911
In summary, Calico versions greater than or equal to 3.19.0 need extra time to finish spinning up, but the default value for the readiness and liveness probes is only 1 second. Even a little extra time would result in these pods succeeding their liveness checks and coming up successfully. The default timeout value was increased to 10 seconds in calico 3.20 which remediates the issue, but this requires upgrading to DKP 2.X in order to take advantage of.
If you are stuck on Konvoy 1.8 at the moment, you can easily patch the value of livenessProbe and readinessProbe's timeoutSeconds to 10 seconds via the following command:
kubectl patch daemonset calico-node -n kube-system -p='{"spec":{"template":{"spec":{"containers":[{"name":"calico-node","livenessProbe":{"timeoutSeconds":10},"readinessProbe":{"timeoutSeconds":10}}]}}}}'
This will immediately restart all calico pods in the daemonset with the new values. You can check the current values via:
kubectl get daemonset -n kube-system calico-node -o yaml