Troubleshoot OpenShift deployment issues

Learn how you can troubleshoot various issues when you deploy MOSTLY AI to an OpenShift cluster. Each of the listed issues includes a description of the problem and solution that shows how to overcome or workaround the issue.

Cannot list or patch resource

Problem

During the deployment of MOSTLY AI to an OpenShift cluster, you might see one of the errors listed below.

Error from server (Forbidden): nodes is forbidden: User "hello@mostly.ai"
cannot list resource "nodes" in API group "" at the cluster scope

Error from server (Forbidden): namespaces "mostly-ai" is forbidden: 
User "hello@mostly.ai" cannot patch resource "namespaces" in
API group "" in the namespace "mostly-ai"

The errors appear, when you attempt to run the commands listed below from the OpenShift deployment guide.

oc get nodes

oc annotate --overwrite namespace mostly-ai openshift.io/sa.scc.uid-range="1000700000/10000"

Solution

Request elevated permissions for your user from your OpenShift administrator.

Violate PodSecurity

The installation error would violate PodSecurity allowPrivilegeEscalation might indicate misconfigurations in the values.yaml file.

Problem

When you run the helm command to install MOSTLY AI on an OpenShift cluster, you might see the installation error similar to the one below.

W0714 14:04:39.656905   47432 warnings.go:70] would violate PodSecurity "restricted:v1.24": allowPrivilegeEscalation != false (container "mostly-*" must set securityContext.allowPrivilegeEscalation=false), unrestricted capabilities (container "mostly-*" must set securityContext.capabilities.drop=["ALL"]), runAsNonRoot != true (pod or container "mostly-*" must set securityContext.runAsNonRoot=true), seccompProfile (pod or container "mostly-*" must set securityContext.seccompProfile.type to "RuntimeDefault" or "Localhost")
W0714 14:04:39.711306   47432 warnings.go:70] would violate PodSecurity "restricted:v1.24": allowPrivilegeEscalation != false (container "mostly-psql" must set securityContext.allowPrivilegeEscalation=false), unrestricted capabilities (container "mostly-psql" must set securityContext.capabilities.drop=["ALL"])

Solution

Verify that you have the correct storage class defined for the CORDINATOR: pvc: storageClassName key. To check your OpenShift storageClassNames, open your OpenShift console, and navigate to Storage > StorageClasses. The name defined for CORDINATOR: pvc: storageClassName must match one of the entires.
Verify that you the platform key is set to ocp and not k8s.
values.yaml
```
platform: k8s # incorrect
platform: ocp # correct
```
Alternatively, keep in mind that this warning reflects the security constraints in your cluster. To better adapt to the, set the platform key to other. Like so, the suggested security contexts of the helm charts will be ignored and the deployment will follow the ones from your cluster.
values.yaml
```
platform: other # alternative
```

Failed to create resource

The error UPGRADE FAILED: failed to create resource: Secret in version "v1" cannot be handled as a Secret might indicate an incorrect or missing secret key in the docker_secret key in the values.yaml file.

Problem

When you run the helm command to deploy MOSTLY AI in an OpenShift cluster, you might see the deployment error similar to the one below.

UPGRADE FAILED: failed to create resource: Secret in version "v1" cannot be handled as a Secret: illegal base64 data at input byte 6

Make sure that you have set the Docker pull image secret in docker_secret.

Solution

Double-check that you have the correct secret key in docker_secret.
Make sure that the secret key value is not wrapped in quotation marks.

If the suggestions above do not help to resolve the issue, contact your Customer Experience Engineer.

Cannot patch "mostly-data" with PersistentVolumeClaim

The error UPGRADE FAILED: cannot patch "mostly-data" with kind PersistentVolumeClaim might indicate intermittent installation issues.

Problem

When you run the helm command to deploy MOSTLY AI in an OpenShift cluster, you might see an installation error similar to the one below.

Error: UPGRADE FAILED: cannot patch "mostly-data" with kind PersistentVolumeClaim: PersistentVolumeClaim "mostly-data" is invalid: spec: Forbidden: spec is immutable after creation except resources.requests for bound claims
  core.PersistentVolumeClaimSpec{
  	... // 2 identical fields
  	Resources:        {Requests: {s"storage": {i: {...}, s: "50Gi", Format: "BinarySI"}}},
  	VolumeName:       "",
- 	StorageClassName: &"efs-sc",
+ 	StorageClassName: &"efs",
  	VolumeMode:       &"Filesystem",
  	DataSource:       nil,
  	DataSourceRef:    nil,
  }

Solution

You can work around the issue above by removing the mostly-ai project from your OpenShift cluster, recreating it, and running the installation again.

Delete the the mostly-ai project from your OpenShift cluster.
```
oc delete namespace mostly-ai
```

Create the mostly-ai project again.

oc new-project mostly-ai

Enter the following annotations for the pods of MOSTLY AI to be able to be scheduled in this namespace:

oc annotate --overwrite namespace mostly-ai openshift.io/sa.scc.supplemental-groups="1000700000/10000"

oc annotate --overwrite namespace mostly-ai openshift.io/sa.scc.uid-range="1000700000/10000"

Run the deployment again with the helm command.

helm upgrade --install mostly-ai ./ --values values.yaml --namespace mostly-ai