Creation known issues topic

content/docs/reference/known-issues.md

@@ -0,0 +1,55 @@
+# Known issues
+
+Workaround solutions are available for the following MKE 4 Alpha.1 known
+issues:
+
+## [BOP-694] MKE 3 upgrade rollback destabilizes the cluster
+
+When a rollback to MKE 3 occurs as the result of an upgrade issue, the MKE
+cluster is inaccessible through the MKE web UI address and ``ucp-auth``
+containers continually restart.
+
+**Workaround:** Uninstall MKE 3 and restore it from a backup.
+
+## [BOP-708] OIDC authenticaion fails after mkectl upgrade
+
+Due to an issue with client secret migration, OIDC authentication fails
+following an upgrade performed with mkectl.
+
+**Workaround:**
+
+1. Copy the MKE 4 config that prints at the end of migration.
+
+2. Update the ``authentication.oidc.clientSecret`` field to the secret field
+   from your identity provider.
+
+3. Apply the updated MKE 4 config.
+
+## [BOP-686] In MKE 3 upgrade, kubectl commands return ``No agent available``
+
+For a cluster with multiple controller nodes (manager/master nodes), [k0s
+requires the presence ofa load balancer for the controller
+node](https://docs.k0sproject.io/head/high-availability/ ). Without a load
+balancer, the controller nodes is unable to reach the kubelet on the worker
+nodes, and thus the user will be presented with ``No agent available`` errors.
+
+**Workaround:**
+
+1. If an external load balancer is not already in place, create one that
+   targets all controllers and that forwards the following ports:
+
+   - `443`, for controller
+   - `6443`, for Kubernetes API
+   - `8132`, for Konnectivity
+
+2. Use `k0sctl` to update the `k0s` config to set `externalAddress`:
+
+   ```
+   k0s:
+     config:
+       spec:
+         api:
+           externalAddress: <load balancer public ip address>
+           sans:
+           - <load balancer public ip address>
+   ```