Add Support for Azure WAL-G Backups (#1537)

This commit adds support for using an Azure storage account as a backup
location.

It uses the existing GCS functionality as a reference for what to do,
and follows the example set by GCS as closely as possible.

The decision to name the cloud provider key "aws_or_gcp" is unfortunate
while adding support for Azure, but I have left it alone to allow for
this changeset to be backwards compatible.

charts/postgres-operator/crds/operatorconfigurations.yaml

@@ -395,6 +395,8 @@ spec:
                     type: string
                   wal_s3_bucket:
                     type: string
+                  wal_az_storage_account:
+                    type: string
               logical_backup:
                 type: object
                 properties:

charts/postgres-operator/values.yaml

@@ -268,6 +268,9 @@ configAwsOrGcp:
   # GCS bucket to use for shipping WAL segments with WAL-E
   # wal_gs_bucket: ""
 
+  # Azure Storage Account to use for shipping WAL segments with WAL-G
+  # wal_az_storage_account: ""
+
 # configure K8s cron job managed by the operator
 configLogicalBackup:
   # image for pods of the logical backup job (example runs pg_dumpall)

docs/administrator.md

@@ -808,6 +808,63 @@ pod_environment_configmap: "postgres-operator-system/pod-env-overrides"
 ...
 ```
 
+### Azure setup
+
+To configure the operator on Azure these prerequisites are needed:
+
+* A storage account in the same region as the Kubernetes cluster.
+
+The configuration parameters that we will be using are:
+
+* `pod_environment_secret`
+* `wal_az_storage_account`
+
+1. Generate the K8s secret resource that will contain your storage account's
+access key. You will need a copy of this secret in every namespace you want to
+create postgresql clusters.
+
+The latest version of WAL-G (v1.0) supports the use of a SASS token, but you'll
+have to make due with using the primary or secondary access token until the
+version of WAL-G is updated in the postgres-operator.
+
+```yaml
+apiVersion: v1
+kind: Secret
+metadata:
+  name: psql-backup-creds
+  namespace: default
+type: Opaque
+stringData:
+  AZURE_STORAGE_ACCESS_KEY: <primary or secondary access key>
+```
+
+2. Setup pod environment configmap that instructs the operator to use WAL-G,
+instead of WAL-E, for backup and restore.
+```yml
+apiVersion: v1
+kind: ConfigMap
+metadata:
+  name: pod-env-overrides
+  namespace: postgres-operator-system
+data:
+  # Any env variable used by spilo can be added
+  USE_WALG_BACKUP: "true"
+  USE_WALG_RESTORE: "true"
+  CLONE_USE_WALG_RESTORE: "true"
+```
+
+3. Setup your operator configuration values. With the `psql-backup-creds`
+and `pod-env-overrides` resources applied to your cluster, ensure that the operator's configuration
+is set up like the following:
+```yml
+...
+aws_or_gcp:
+  pod_environment_secret: "pgsql-backup-creds"
+  pod_environment_configmap: "postgres-operator-system/pod-env-overrides"
+  wal_az_storage_account: "postgresbackupsbucket28302F2"  # name of storage account to save the WAL-G logs
+...
+```
+
 ### Restoring physical backups
 
 If cluster members have to be (re)initialized restoring physical backups

docs/reference/operator_parameters.md

@@ -557,6 +557,12 @@ yet officially supported.
   [service accounts](https://cloud.google.com/kubernetes-engine/docs/tutorials/authenticating-to-cloud-platform).
   The default is empty
 
+* **wal_az_storage_account**
+  Azure Storage Account to use for shipping WAL segments with WAL-G. The
+  storage account must exist and be accessible by Postgres pods. Note, only the
+  name of the storage account is required.
+  The default is empty.
+
 * **log_s3_bucket**
   S3 bucket to use for shipping Postgres daily logs. Works only with S3 on AWS.
   The bucket has to be present and accessible by Postgres pods. The default is

manifests/configmap.yaml

@@ -129,6 +129,7 @@ data:
   # team_api_role_configuration: "log_statement:all"
   # teams_api_url: http://fake-teams-api.default.svc.cluster.local
   # toleration: ""
+  # wal_az_storage_account: ""
   # wal_gs_bucket: ""
   # wal_s3_bucket: ""
   watched_namespace: "*"  # listen to all namespaces

manifests/operatorconfiguration.crd.yaml

@@ -384,6 +384,8 @@ spec:
                     type: string
                   log_s3_bucket:
                     type: string
+                  wal_az_storage_account:
+                    type: string
                   wal_gs_bucket:
                     type: string
                   wal_s3_bucket:

manifests/postgresql-operator-default-configuration.yaml

@@ -121,6 +121,7 @@ configuration:
     # gcp_credentials: ""
     # kube_iam_role: ""
     # log_s3_bucket: ""
+    # wal_az_storage_account: ""
     # wal_gs_bucket: ""
     # wal_s3_bucket: ""
   logical_backup:

pkg/apis/acid.zalan.do/v1/operator_configuration_type.go

@@ -132,6 +132,7 @@ type AWSGCPConfiguration struct {
 	AWSRegion                    string `json:"aws_region,omitempty"`
 	WALGSBucket                  string `json:"wal_gs_bucket,omitempty"`
 	GCPCredentials               string `json:"gcp_credentials,omitempty"`
+	WALAZStorageAccount          string `json:"wal_az_storage_account,omitempty"`
 	LogS3Bucket                  string `json:"log_s3_bucket,omitempty"`
 	KubeIAMRole                  string `json:"kube_iam_role,omitempty"`
 	AdditionalSecretMount        string `json:"additional_secret_mount,omitempty"`

pkg/cluster/k8sres.go

@@ -798,6 +798,12 @@ func (c *Cluster) generateSpiloPodEnvVars(uid types.UID, spiloConfiguration stri
 		envVars = append(envVars, v1.EnvVar{Name: "WAL_BUCKET_SCOPE_PREFIX", Value: ""})
 	}
 
+	if c.OpConfig.WALAZStorageAccount != "" {
+		envVars = append(envVars, v1.EnvVar{Name: "AZURE_STORAGE_ACCOUNT", Value: c.OpConfig.WALAZStorageAccount})
+		envVars = append(envVars, v1.EnvVar{Name: "WAL_BUCKET_SCOPE_SUFFIX", Value: getBucketScopeSuffix(string(uid))})
+		envVars = append(envVars, v1.EnvVar{Name: "WAL_BUCKET_SCOPE_PREFIX", Value: ""})
+	}
+
 	if c.OpConfig.GCPCredentials != "" {
 		envVars = append(envVars, v1.EnvVar{Name: "GOOGLE_APPLICATION_CREDENTIALS", Value: c.OpConfig.GCPCredentials})
 	}
@@ -1805,6 +1811,14 @@ func (c *Cluster) generateCloneEnvironment(description *acidv1.CloneDescription)
 					},
 				}
 				result = append(result, envs...)
+			} else if c.OpConfig.WALAZStorageAccount != "" {
+				envs := []v1.EnvVar{
+					{
+						Name:  "CLONE_AZURE_STORAGE_ACCOUNT",
+						Value: c.OpConfig.WALAZStorageAccount,
+					},
+				}
+				result = append(result, envs...)
 			} else {
 				c.logger.Error("Cannot figure out S3 or GS bucket. Both are empty.")
 			}

pkg/controller/operator_config.go

@@ -146,6 +146,7 @@ func (c *Controller) importConfigurationFromCRD(fromCRD *acidv1.OperatorConfigur
 	result.KubeIAMRole = fromCRD.AWSGCP.KubeIAMRole
 	result.WALGSBucket = fromCRD.AWSGCP.WALGSBucket
 	result.GCPCredentials = fromCRD.AWSGCP.GCPCredentials
+	result.WALAZStorageAccount = fromCRD.AWSGCP.WALAZStorageAccount
 	result.AdditionalSecretMount = fromCRD.AWSGCP.AdditionalSecretMount
 	result.AdditionalSecretMountPath = util.Coalesce(fromCRD.AWSGCP.AdditionalSecretMountPath, "/meta/credentials")
 	result.EnableEBSGp3Migration = fromCRD.AWSGCP.EnableEBSGp3Migration

pkg/util/config/config.go

@@ -167,6 +167,7 @@ type Config struct {
 	KubeIAMRole                            string            `name:"kube_iam_role"`
 	WALGSBucket                            string            `name:"wal_gs_bucket"`
 	GCPCredentials                         string            `name:"gcp_credentials"`
+	WALAZStorageAccount                    string            `name:"wal_az_storage_account"`
 	AdditionalSecretMount                  string            `name:"additional_secret_mount"`
 	AdditionalSecretMountPath              string            `name:"additional_secret_mount_path" default:"/meta/credentials"`
 	EnableEBSGp3Migration                  bool              `name:"enable_ebs_gp3_migration" default:"false"`