Known issues

The MKE 4 known issues with available workarounds are described herein.

[BOP-583] LDAP settings fail to migrate during upgrade from MKE 3

LDAP configurations are not stored in MKE 3 configuration files, and thus they are not included when you upgrade to MKE 4 from an MKE 3 installation.

Workaround:

When upgrading from MKE 3, you must manually add the LDAP configuration.

1. Make a request to https://{{host}}/enzi/v0/config/auth/ldap on the MKE 3 cluster prior to the migration. For more information, refer to the MKE 3 LDAP Configuration through API documentation.

2. Convert the LDAP response to the MKE 4 LDAP settings.

3. Apply the translated LDAP settings to the cluster following migration.

[BOP-708] OIDC authentication 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 configuration 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 configuration.

[BOP-898][BOP-899] Calico eBPF and IPVS modes are not supported

Calico eBPF and IPVS mode are not yet supported for MKE 4. As such, upgrading from an MKE 3 cluster using either of those networking modes results in an error:

FATA[0640] Upgrade failed due to error: failed to run step [Upgrade Tasks]:
unable to install BOP: unable to apply MKE4 config: failed to wait for pods:
failed to wait for pods: failed to list pods: client rate limiter Wait returned
an error: context deadline exceeded

[BOP-905] Prometheus dashboard reports incorrect heavy memory use

The Prometheus dashboard displays heavy memory use that does not accurately reflect true memory status.

[BOP-947] Managed user passwords are not migrated during upgrade from MKE 3

The admin password is migrated during upgrade from MKE 3, however all other managed user passwords are not migrated.

[BOP-964] mke-operator in crashloopbackoff status

The mke-operator-controller-manager is in crashloopbackoff status in MKE 4 Alpha 2. You can safely ignore this, however, as it has no effect on MKE 4.0.0-alpha.2.0 functionality.

[BOP-982] Cannot change MKE 4 password using mkectl

You cannot change the password for an existing MKE 4 deployment by running mkectl apply -f mke4.yaml --admin-password <password>.

Workaround:

Use kubectl to change the Password object:

1. Obtain the list of users:

   $ kubectl -n mke get passwords -o custom-columns=NAME:.metadata.name,EMAIL:.email

   Example output:

   NAME                    EMAIL
   mfsg22lozpzjzzeeeirsk   admin

2. Reveal the Password object for the target user.

   $ kubectl -n mke get password mfsg22lozpzjzzeeeirsk -oyaml

   Example output:

   apiVersion: dex.coreos.com/v1
   email: admin
   hash: JDJhJDEwJFA5RUppWmVJLkRCMVlqMWJqZk5rUk9RQ1oybFFpOUhXUFhnYmIxdUFPSkpHeGFDWUl1OTcy
   kind: Password
   metadata:
     creationTimestamp: "2024-07-23T18:39:11Z"
     generation: 1
     name: mfsg22lozpzjzzeeeirsk
     namespace: mke
     resourceVersion: "3558"
     uid: 91a9e728-abfa-4daa-bdab-4c09cf888281
   userID: 7668fdb9-a979-4645-b6cc-10985df77da6
   username: admin

3. Edit the hash field with the desired password hash.

[BOP-1299] Disk Usage and Memory metrics are not shown correctly in the dashboard with disabled cAdvisor

When cAdvisor is disabled, the main page of the dashboard presents 0% as the value for the Disk Usage and Memory metrics.

Workaround:

Enable cAdvisor in the MKE configuration file and run mkectl apply.

monitoring:
  enableCAdvisor: true

[BOP-1299] Max Used Disk and Max CPU labels are swapped in the MKE dashboard

No workaround is available at this time.

[BOP-1307] Prometheus can be accessed without authentication

Any party with knowledge of the MKE 4 URL can access Prometheus without authentication.

No workaround is available at this time.

[BOP-891] Upgrade to MKE 4 fails if kubeconfig file is present in source MKE 3.x

Upgrade to MKE 4 fails if the ~/.mke/mke.kubeconf file is present in the source MKE 3.x system.

Workaround:

Make a backup of the old ~/.mke/mke.kubeconf file and then delete it.

[BOP-1528] Once applied, the apiserver.externalAddress parameter cannot be cleared

MKE cannot clear the apiserver.externalAddress parameter once it has been applied in the MKE 4 configuration file, as this can cause the MKE cluster to malfunction.

No workaround is available at this time.