How exactly does it restart? Do you see that line repeated It should loop until success, exit the init container once that open line appears, and then proceed with the main container, e.g.
kubectl logs example-kong-init-migrations-x7886 -c wait-for-postgres
nc: example-postgresql (10.19.245.143:5432): Connection timed out
waiting for db - trying example-postgresql:5432
nc: example-postgresql (10.19.245.143:5432): Connection timed out
waiting for db - trying example-postgresql:5432
example-postgresql (10.19.245.143:5432) open
The open line should never appear more than once, and you should see that the init container stops after it appears, and that you can run kubectl logs PODNAME -c kong-migrations after to see their progress (or any failures beyond a basic connection failure).
If the pod is in fact restarting, do you see anything in the kubectl describe events output for the pod or job? Kubernetes should log any external reasons for a pod restart, although we don’t define a deadline or other reasons for killing the pod–I suspect what you’re running into now is that the main migrations container is starting and exiting unsuccessfully for some other reason, e.g. bad auth credentials or missing database permissions.