Chapter 1. Backing up applications
You can employ OpenShift API for Data Protection (OADP) with Red Hat OpenShift Service on AWS (ROSA) clusters to backup and restore application data. Before installing OADP, you must set up role and policy credentials for OADP so that it can use the AWS API.
This is a two stage process:
- Prepare AWS credentials.
- Install the OADP Operator and provide it with the IAM role.
1.1. Preparing AWS credentials Copy linkLink copied to clipboard!
An AWS account must be ready to accept an OADP installation.
Procedure
Create the following environment variables by running the following commands:
NoteChange the cluster name to match your ROSA cluster, and ensure you are logged into the cluster as an administrator. Ensure that all fields are outputted correctly before continuing.
$ export CLUSTER_NAME=my-cluster1 export ROSA_CLUSTER_ID=$(rosa describe cluster -c ${CLUSTER_NAME} --output json | jq -r .id) export REGION=$(rosa describe cluster -c ${CLUSTER_NAME} --output json | jq -r .region.id) export OIDC_ENDPOINT=$(oc get authentication.config.openshift.io cluster -o jsonpath='{.spec.serviceAccountIssuer}' | sed 's|^https://||') export AWS_ACCOUNT_ID=$(aws sts get-caller-identity --query Account --output text) export CLUSTER_VERSION=$(rosa describe cluster -c ${CLUSTER_NAME} -o json | jq -r .version.raw_id | cut -f -2 -d '.') export ROLE_NAME="${CLUSTER_NAME}-openshift-oadp-aws-cloud-credentials" export SCRATCH="/tmp/${CLUSTER_NAME}/oadp" mkdir -p ${SCRATCH} echo "Cluster ID: ${ROSA_CLUSTER_ID}, Region: ${REGION}, OIDC Endpoint: ${OIDC_ENDPOINT}, AWS Account ID: ${AWS_ACCOUNT_ID}"- 1
- Replace
my-clusterwith your ROSA cluster name.
On the AWS account, create an IAM policy to allow access to S3.
Check to see if the policy exists by running the following command:
$ POLICY_ARN=$(aws iam list-policies --query "Policies[?PolicyName=='RosaOadpVer1'].{ARN:Arn}" --output text)1 - 1
- Replace
RosaOadpwith your policy name.
Use the following command to create the policy JSON file and then create the policy in ROSA.
NoteIf the policy ARN is not found, the command will create the policy. If the policy ARN already exists, the
ifstatement will intentionally skip the policy creation.$ if [[ -z "${POLICY_ARN}" ]]; then cat << EOF > ${SCRATCH}/policy.json1 { "Version": "2012-10-17", "Statement": [ { "Effect": "Allow", "Action": [ "s3:CreateBucket", "s3:DeleteBucket", "s3:PutBucketTagging", "s3:GetBucketTagging", "s3:PutEncryptionConfiguration", "s3:GetEncryptionConfiguration", "s3:PutLifecycleConfiguration", "s3:GetLifecycleConfiguration", "s3:GetBucketLocation", "s3:ListBucket", "s3:GetObject", "s3:PutObject", "s3:DeleteObject", "s3:ListBucketMultipartUploads", "s3:AbortMultipartUpload", "s3:ListMultipartUploadParts", "ec2:DescribeSnapshots", "ec2:DescribeVolumes", "ec2:DescribeVolumeAttribute", "ec2:DescribeVolumesModifications", "ec2:DescribeVolumeStatus", "ec2:CreateTags", "ec2:CreateVolume", "ec2:CreateSnapshot", "ec2:DeleteSnapshot" ], "Resource": "*" } ]} EOF POLICY_ARN=$(aws iam create-policy --policy-name "RosaOadpVer1" \ --policy-document file:///${SCRATCH}/policy.json --query Policy.Arn \ --tags Key=rosa_openshift_version,Value=${CLUSTER_VERSION} Key=rosa_role_prefix,Value=ManagedOpenShift Key=operator_namespace,Value=openshift-oadp Key=operator_name,Value=openshift-oadp \ --output text) fi- 1
SCRATCHis a name for a temporary directory created for the environment variables.
View the policy ARN by running the following command:
$ echo ${POLICY_ARN}
Create an IAM role trust policy for the cluster:
Create the trust policy file by running the following command:
$ cat <<EOF > ${SCRATCH}/trust-policy.json { "Version": "2012-10-17", "Statement": [{ "Effect": "Allow", "Principal": { "Federated": "arn:aws:iam::${AWS_ACCOUNT_ID}:oidc-provider/${OIDC_ENDPOINT}" }, "Action": "sts:AssumeRoleWithWebIdentity", "Condition": { "StringEquals": { "${OIDC_ENDPOINT}:sub": [ "system:serviceaccount:openshift-adp:openshift-adp-controller-manager", "system:serviceaccount:openshift-adp:velero"] } } }] } EOFCreate the role by running the following command:
$ ROLE_ARN=$(aws iam create-role --role-name \ "${ROLE_NAME}" \ --assume-role-policy-document file://${SCRATCH}/trust-policy.json \ --tags Key=rosa_cluster_id,Value=${ROSA_CLUSTER_ID} Key=rosa_openshift_version,Value=${CLUSTER_VERSION} Key=rosa_role_prefix,Value=ManagedOpenShift Key=operator_namespace,Value=openshift-adp Key=operator_name,Value=openshift-oadp \ --query Role.Arn --output text)View the role ARN by running the following command:
$ echo ${ROLE_ARN}
Attach the IAM policy to the IAM role by running the following command:
$ aws iam attach-role-policy --role-name "${ROLE_NAME}" \ --policy-arn ${POLICY_ARN}
Next steps
- Continue to Installing the OADP Operator and providing the IAM role.
1.2. Installing the OADP Operator and providing the IAM role Copy linkLink copied to clipboard!
AWS Security Token Service (AWS STS) is a global web service that provides short-term credentials for IAM or federated users. Red Hat OpenShift Service on AWS (ROSA) with STS is the recommended credential mode for ROSA clusters. This document describes how to install OpenShift API for Data Protection (OADP) on ROSA with AWS STS.
Restic and Kopia are not supported in the OADP on ROSA with AWS STS environment. Verify that the Restic and Kopia node agent is disabled. For backing up volumes, OADP on ROSA with AWS STS supports only native snapshots and Container Storage Interface (CSI) snapshots.
In an Amazon ROSA cluster that uses STS authentication, restoring backed-up data in a different AWS region is not supported.
The Data Mover feature is not currently supported in ROSA clusters. You can use native AWS S3 tools for moving data.
Prerequisites
-
An Red Hat OpenShift Service on AWS ROSA cluster with the required access and tokens. For instructions, see the previous procedure Preparing AWS credentials for OADP. If you plan to use two different clusters for backing up and restoring, you must prepare AWS credentials, including
ROLE_ARN, for each cluster.
Procedure
Create an Red Hat OpenShift Service on AWS secret from your AWS token file by entering the following commands:
Create the credentials file:
$ cat <<EOF > ${SCRATCH}/credentials [default] role_arn = ${ROLE_ARN} web_identity_token_file = /var/run/secrets/openshift/serviceaccount/token EOFCreate a namespace for OADP:
$ oc create namespace openshift-adpCreate the Red Hat OpenShift Service on AWS secret:
$ oc -n openshift-adp create secret generic cloud-credentials \ --from-file=${SCRATCH}/credentialsNoteIn Red Hat OpenShift Service on AWS versions 4.15 and later, the OADP Operator supports a new standardized STS workflow through the Operator Lifecycle Manager (OLM) and Cloud Credentials Operator (CCO). In this workflow, you do not need to create the above secret, you only need to supply the role ARN during the installation of OLM-managed operators using the Red Hat OpenShift Service on AWS web console, for more information see Installing from OperatorHub using the web console.
The preceding secret is created automatically by CCO.
Install the OADP Operator:
-
In the Red Hat OpenShift Service on AWS web console, browse to Operators
OperatorHub. - Search for the OADP Operator.
- In the role_ARN field, paste the role_arn that you created previously and click Install.
-
In the Red Hat OpenShift Service on AWS web console, browse to Operators
Create AWS cloud storage using your AWS credentials by entering the following command:
$ cat << EOF | oc create -f - apiVersion: oadp.openshift.io/v1alpha1 kind: CloudStorage metadata: name: ${CLUSTER_NAME}-oadp namespace: openshift-adp spec: creationSecret: key: credentials name: cloud-credentials enableSharedConfig: true name: ${CLUSTER_NAME}-oadp provider: aws region: $REGION EOFCheck your application’s storage default storage class by entering the following command:
$ oc get pvc -n <namespace>Example output
NAME STATUS VOLUME CAPACITY ACCESS MODES STORAGECLASS AGE applog Bound pvc-351791ae-b6ab-4e8b-88a4-30f73caf5ef8 1Gi RWO gp3-csi 4d19h mysql Bound pvc-16b8e009-a20a-4379-accc-bc81fedd0621 1Gi RWO gp3-csi 4d19hGet the storage class by running the following command:
$ oc get storageclassExample output
NAME PROVISIONER RECLAIMPOLICY VOLUMEBINDINGMODE ALLOWVOLUMEEXPANSION AGE gp2 kubernetes.io/aws-ebs Delete WaitForFirstConsumer true 4d21h gp2-csi ebs.csi.aws.com Delete WaitForFirstConsumer true 4d21h gp3 ebs.csi.aws.com Delete WaitForFirstConsumer true 4d21h gp3-csi (default) ebs.csi.aws.com Delete WaitForFirstConsumer true 4d21hNoteThe following storage classes will work:
- gp3-csi
- gp2-csi
- gp3
- gp2
If the application or applications that are being backed up are all using persistent volumes (PVs) with Container Storage Interface (CSI), it is advisable to include the CSI plugin in the OADP DPA configuration.
Create the
DataProtectionApplicationresource to configure the connection to the storage where the backups and volume snapshots are stored:If you are using only CSI volumes, deploy a Data Protection Application by entering the following command:
$ cat << EOF | oc create -f - apiVersion: oadp.openshift.io/v1alpha1 kind: DataProtectionApplication metadata: name: ${CLUSTER_NAME}-dpa namespace: openshift-adp spec: backupImages: true1 features: dataMover: enable: false backupLocations: - bucket: cloudStorageRef: name: ${CLUSTER_NAME}-oadp credential: key: credentials name: cloud-credentials prefix: velero default: true config: region: ${REGION} configuration: velero: defaultPlugins: - openshift - aws - csi restic: enable: false EOF- 1
- ROSA supports internal image backup. Set this field to
falseif you do not want to use image backup.
If you are using CSI or non-CSI volumes, deploy a Data Protection Application by entering the following command:
$ cat << EOF | oc create -f - apiVersion: oadp.openshift.io/v1alpha1 kind: DataProtectionApplication metadata: name: ${CLUSTER_NAME}-dpa namespace: openshift-adp spec: backupImages: true1 features: dataMover: enable: false backupLocations: - bucket: cloudStorageRef: name: ${CLUSTER_NAME}-oadp credential: key: credentials name: cloud-credentials prefix: velero default: true config: region: ${REGION} configuration: velero: defaultPlugins: - openshift - aws nodeAgent:2 enable: false uploaderType: restic snapshotLocations: - velero: config: credentialsFile: /tmp/credentials/openshift-adp/cloud-credentials-credentials3 enableSharedConfig: "true"4 profile: default5 region: ${REGION}6 provider: aws EOF- 1
- ROSA supports internal image backup. Set this field to false if you do not want to use image backup.
- 2
- See the important note regarding the
nodeAgentattribute. - 3
- The
credentialsFilefield is the mounted location of the bucket credential on the pod. - 4
- The
enableSharedConfigfield allows thesnapshotLocationsto share or reuse the credential defined for the bucket. - 5
- Use the profile name set in the AWS credentials file.
- 6
- Specify
regionas your AWS region. This must be the same as the cluster region.
You are now ready to back up and restore Red Hat OpenShift Service on AWS applications, as described in Backing up applications.
The enable parameter of restic is set to false in this configuration, because OADP does not support Restic in ROSA environments.
If you use OADP 1.2, replace this configuration:
nodeAgent:
enable: false
uploaderType: restic
with the following configuration:
restic:
enable: false
If you want to use two different clusters for backing up and restoring, the two clusters must have the same AWS S3 storage names in both the cloud storage CR and the OADP DataProtectionApplication configuration.
1.3. Backing up workloads on OADP with ROSA STS Copy linkLink copied to clipboard!
1.3.1. Performing a backup with OADP and ROSA STS Copy linkLink copied to clipboard!
The following example hello-world application has no persistent volumes (PVs) attached. Perform a backup with OpenShift API for Data Protection (OADP) with Red Hat OpenShift Service on AWS (ROSA) STS.
Either Data Protection Application (DPA) configuration will work.
Create a workload to back up by running the following commands:
$ oc create namespace hello-world$ oc new-app -n hello-world --image=docker.io/openshift/hello-openshiftExpose the route by running the following command:
$ oc expose service/hello-openshift -n hello-worldCheck that the application is working by running the following command:
$ curl `oc get route/hello-openshift -n hello-world -o jsonpath='{.spec.host}'`Example output
Hello OpenShift!Back up the workload by running the following command:
$ cat << EOF | oc create -f - apiVersion: velero.io/v1 kind: Backup metadata: name: hello-world namespace: openshift-adp spec: includedNamespaces: - hello-world storageLocation: ${CLUSTER_NAME}-dpa-1 ttl: 720h0m0s EOFWait until the backup is completed and then run the following command:
$ watch "oc -n openshift-adp get backup hello-world -o json | jq .status"Example output
{ "completionTimestamp": "2022-09-07T22:20:44Z", "expiration": "2022-10-07T22:20:22Z", "formatVersion": "1.1.0", "phase": "Completed", "progress": { "itemsBackedUp": 58, "totalItems": 58 }, "startTimestamp": "2022-09-07T22:20:22Z", "version": 1 }Delete the demo workload by running the following command:
$ oc delete ns hello-worldRestore the workload from the backup by running the following command:
$ cat << EOF | oc create -f - apiVersion: velero.io/v1 kind: Restore metadata: name: hello-world namespace: openshift-adp spec: backupName: hello-world EOFWait for the Restore to finish by running the following command:
$ watch "oc -n openshift-adp get restore hello-world -o json | jq .status"Example output
{ "completionTimestamp": "2022-09-07T22:25:47Z", "phase": "Completed", "progress": { "itemsRestored": 38, "totalItems": 38 }, "startTimestamp": "2022-09-07T22:25:28Z", "warnings": 9 }Check that the workload is restored by running the following command:
$ oc -n hello-world get podsExample output
NAME READY STATUS RESTARTS AGE hello-openshift-9f885f7c6-kdjpj 1/1 Running 0 90sCheck the JSONPath by running the following command:
$ curl `oc get route/hello-openshift -n hello-world -o jsonpath='{.spec.host}'`Example output
Hello OpenShift!
For troubleshooting tips, see the OADP team’s troubleshooting documentation.
1.3.2. Cleaning up a cluster after a backup with OADP and ROSA STS Copy linkLink copied to clipboard!
If you need to uninstall the OpenShift API for Data Protection (OADP) Operator together with the backups and the S3 bucket from this example, follow these instructions.
Procedure
Delete the workload by running the following command:
$ oc delete ns hello-worldDelete the Data Protection Application (DPA) by running the following command:
$ oc -n openshift-adp delete dpa ${CLUSTER_NAME}-dpaDelete the cloud storage by running the following command:
$ oc -n openshift-adp delete cloudstorage ${CLUSTER_NAME}-oadpWarningIf this command hangs, you might need to delete the finalizer by running the following command:
$ oc -n openshift-adp patch cloudstorage ${CLUSTER_NAME}-oadp -p '{"metadata":{"finalizers":null}}' --type=mergeIf the Operator is no longer required, remove it by running the following command:
$ oc -n openshift-adp delete subscription oadp-operatorRemove the namespace from the Operator:
$ oc delete ns openshift-adpIf the backup and restore resources are no longer required, remove them from the cluster by running the following command:
$ oc delete backup hello-worldTo delete backup, restore and remote objects in AWS S3 run the following command:
$ velero backup delete hello-worldIf you no longer need the Custom Resource Definitions (CRD), remove them from the cluster by running the following command:
$ for CRD in `oc get crds | grep velero | awk '{print $1}'`; do oc delete crd $CRD; doneDelete the AWS S3 bucket by running the following commands:
$ aws s3 rm s3://${CLUSTER_NAME}-oadp --recursive$ aws s3api delete-bucket --bucket ${CLUSTER_NAME}-oadpDetach the policy from the role by running the following command:
$ aws iam detach-role-policy --role-name "${ROLE_NAME}" --policy-arn "${POLICY_ARN}"Delete the role by running the following command:
$ aws iam delete-role --role-name "${ROLE_NAME}"
1.4. Known issues Copy linkLink copied to clipboard!
- Restic, Kopia, and DataMover are not supported or recommended.
- CloudStorage: openshift-adp-controller-manager crashloop seg fault with Restic enabled.
- (Affects OADP 1.1.x_ only): CloudStorage: bucket is removed on CS CR delete, although it doesn’t have "oadp.openshift.io/cloudstorage-delete": "true".