Skip to content

Pulp Operator storage configuration#

Before installing Pulp, for production clusters, it is necessary to configure how Pulp should persist the data.

Pulp uses django-storages to support multiple types of storage backends. The current version of operator supports the following types of storage installation:

Info

Only one storage type should be provided, trying to configure Pulp CR with multiple storage types will fail operator execution.

If no backend is configured, Pulp will by default use the EmptyDir volume.

Configuring Pulp Operator storage to use a Storage Class#

Pulp operator has the following parameters to configure the components with a Storage Class:

  • FileStorageClass - defines the name of the Storage Class that will be used by Pulp core pods
  • Database.PostgresStorageClass - defines the name of the Storage Class that will be used by Database pods
  • Cache.RedisStorageClass - defines the name of the Storage Class that will be used by Cache pods

When Pulp operator is configured with the above parameters it will automatically provision new Persistent Volume Claims with the Storage Class provided.

To verify if there is a Storage Class available: 

$ kubectl get sc

If the Kubernetes cluster has no Storage Class configured, it is possible to configure Pulp with other parameters of storage or follow the steps to create a new Storage Class.

Note

If the Storage Class defined will provision RWO volumes, it is recommended to also set the Deployment strategy in Pulp CR as Recreate to avoid the Multi-Attach volume error.

Configuring Pulp Operator storage to use a Persistent Volume Claim#

Pulp operator has the following parameters to configure the components with a Persistent Volume Claim:

  • PVC - defines the name of the Persistent Volume Claim that will be used by Pulp core pods
  • Database.PVC - defines the name of the Persistent Volume Claim that will be used by Database pods
  • Cache.PVC - defines the name of the Persistent Volume Claim that will be used by Cache pods

When Pulp operator is configured with the above parameters it is expected that the PVCs are already provisioned and Pulp operator will automatically configure the Deployments and StatefulSet with them.

To verify the list of Persistent Volume Claims available: 

$ kubectl get pvc

If the installation namespace has no Persistent Volume Claim available, it is possible to configure Pulp with other parameters of storage or follow the steps to create a new Persistent Volume Claim.

Note

If the Persistent Volume Claim defined is bound to a RWO volume, it is recommended to also set the Deployment strategy in Pulp CR as Recreate to avoid the Multi-Attach volume error.

Configuring Pulp Operator to use object storage#

Pulp operator has the following parameters to configure Pulp core components with Object Storage:

  • ObjectStorageAzureSecret - defines the name of the secret with Azure compliant object storage configuration.
  • ObjectStorageS3Secret - defines the name of the secret with S3 compliant object storage configuration.

When Pulp operator is configured with one of the above parameters it is expected that the secrets are already present in the namespace of Pulp installation. Pulp operator will automatically configure Pulp settings.py with the provided Object Storage backend.

Info

Only one type of Object Storage should be provided. Trying to declare both will fail operator execution.

Configuring Azure Blob#

Prerequisites#

After performing all the prerequisites, create a Secret with them: 

$ PULP_NAMESPACE='my-pulp-namespace'
$ AZURE_ACCOUNT_NAME='my-azure-account-name'
$ AZURE_ACCOUNT_KEY='my-azure-account-key'
$ AZURE_CONTAINER='pulp-test'
$ AZURE_CONTAINER_PATH='pulp3'
$ AZURE_CONNECTION_STRING='my-azure-connection-string'

$ kubectl -n $PULP_NAMESPACE apply -f- <<EOF
apiVersion: v1
kind: Secret
metadata:
  name: 'test-azure'
stringData:
  azure-account-name: $AZURE_ACCOUNT_NAME
  azure-account-key: $AZURE_ACCOUNT_KEY
  azure-container: $AZURE_CONTAINER
  azure-container-path: $AZURE_CONTAINER_PATH
  azure-connection-string: $AZURE_CONNECTION_STRING
EOF

Note

azure-connection-string is an optional field that can be used to keep compatibility with other Azure Storage compliant systems, like Azurite.

Now configure Pulp CR with the secret created: 

$ kubectl -n $PULP_NAMESPACE edit pulp
...
spec:
  object_storage_azure_secret: test-azure
...

and restart the API pods to get the new configuration. 

$ kubectl -n $PULP_NAMESPACE delete pod -l app.kubernetes.io/component=api

Configure AWS S3#

Prerequisites#

  • To configure Pulp with AWS S3 as a storage backend, the first thing to do is create a S3 Bucket to store the objects.
  • After configuring a S3 Bucket take a note of the AWS credentials

After performing all the prerequisites, create a Secret with them: 

$ PULP_NAMESPACE='my-pulp-namespace'
$ S3_ACCESS_KEY_ID='my-aws-access-key-id'
$ S3_SECRET_ACCESS_KEY='my-aws-secret-access-key'
$ S3_BUCKET_NAME='pulp3'
$ S3_REGION='us-east-1'

$ kubectl -n $PULP_NAMESPACE apply -f- <<EOF
apiVersion: v1
kind: Secret
metadata:
  name: 'test-s3'
stringData:
  s3-access-key-id: $S3_ACCESS_KEY_ID
  s3-secret-access-key: $S3_SECRET_ACCESS_KEY
  s3-bucket-name: $S3_BUCKET_NAME
  s3-region: $S3_REGION
EOF

Now configure Pulp CR with the secret created: 

$ kubectl -n $PULP_NAMESPACE edit pulp
...
spec:
  object_storage_s3_secret: test-s3
...

and restart the API pods to get the new configuration. 

$ kubectl -n $PULP_NAMESPACE delete pod -l app.kubernetes.io/component=api

Configuring Pulp Operator in non-production clusters#

If there is no Storage Class nor Persistent Volume Claim nor Object Storage provided the operator will deploy the components (Pulp, Database, and Cache) with an emptyDir.

You must configure storage for the Pulp Operator. For non-production clusters, you can set the components to an empty directory. If you do so, everything is lost if you restart the pod.

Warning

Configure this option for only non-production clusters.

Last update: 2023-03-24