OCI Provider sends many requests to backend in error scenarios

[anders-swanson]

What happened:

When using the OCI Provider in an induced error scenario (specifically with OKE, though may occur in other scenarios), many outbound calls to the backend cloud provider were observed.

These requests were not retryable, e.g., creating a cluster with mismatched Kubernetes versions between the control plane and worker node results in 409 errors from the cloud backend.

What you expected to happen:

• In the case of non-retryable errors (409's and the like), a Condition error is set by the infrastructure provider, and no further retrys are sent. This will avoid unnecessary API calls.
• In the case of retryable errors (5xx, or other), use a backoff mechanism.

How to reproduce it (as minimally and precisely as possible):

Create a managed controlplane with at least one node pool, setting the kubernetes version of the node pool to be different than the control plane.

Anything else we need to know?:

Environment:

• CAPOCI version: 12.1
• Cluster-API version (use clusterctl version):
• Kubernetes version (use kubectl version):
• Docker version (use docker info):
• OS (e.g. from /etc/os-release):

[shyamradhakrishnan]

@anders-swanson the backoff bug was fixed in v0.13.0 via #316