New to KubeDB? Please start here.

Branch Redis Across Clusters

Cross-cluster branching creates an independent Redis deployment on another Kubernetes cluster from a point-in-time snapshot of an existing deployment. Use it to prepare development, testing, or data analysis environments on a separate cluster using existing data.

The branch accepts independent reads and writes without changing the source. Source changes are not continuously replicated; scheduled refreshes can replace branch data with fresh source snapshots.

Before You Begin

  • Read the branching overview.
  • Complete Courier setup, including KubeDB and the Branch CRD. Use branching-capable Courier, provisioner, and database images.
  • Ensure RedisVersion 7.4.0 is available in your installed catalog.
  • Ensure the RedisSentinel CRD is installed; Sentinel uses the same RedisVersion catalog entry. Use Ceph RBD storage with copy-on-write cloning and a compatible VolumeSnapshotClass.
  • Review the snapshot requirements. Your account must be able to create databases and Branches, read auth Secrets, inspect snapshots and Jobs, and open pod shells.
  • Allow capacity for three source Redis pods, three Sentinel pods, the target database, and any action Jobs.
  • Have existing OCM registration for managed clusters source and target, with Courier agents enabled and kubectl contexts hub, source, and target.
  • Both clusters must access the same network-reachable Ceph backend, with compatible CSI identities, pool settings, and credentials. Matching StorageClass names alone do not provide shared snapshots.

The source uses demo; the target uses dev. Replace ceph-rbd consistently if your Ceph StorageClass has another name. Reserve the tutorial names and create namespaces if they do not already exist:

kubectl --context source create namespace demo
kubectl --context target create namespace dev

Check the existing registration and storage classes:

kubectl --context hub get managedclusters source target
kubectl --context source get storageclass ceph-rbd
kubectl --context target get storageclass ceph-rbd

Deploy the source database

Save the following as redis-source.yaml. It creates a three-node Redis source managed by three Sentinels:

apiVersion: kubedb.com/v1
kind: RedisSentinel
metadata:
  name: redis-source-sentinel
  namespace: demo
spec:
  version: 7.4.0
  replicas: 3
  storageType: Durable
  storage:
    storageClassName: ceph-rbd
    accessModes:
    - ReadWriteOnce
    resources:
      requests:
        storage: 1Gi
  deletionPolicy: WipeOut
---
apiVersion: kubedb.com/v1
kind: Redis
metadata:
  name: redis-source
  namespace: demo
spec:
  version: 7.4.0
  replicas: 3
  mode: Sentinel
  sentinelRef:
    name: redis-source-sentinel
    namespace: demo
  podTemplate:
    spec:
      containers:
      - name: redis
        resources:
          requests:
            cpu: 500m
            memory: 1Gi
          limits:
            memory: 1Gi
  storageType: Durable
  storage:
    storageClassName: ceph-rbd
    accessModes:
    - ReadWriteOnce
    resources:
      requests:
        storage: 2Gi
  deletionPolicy: WipeOut
kubectl --context source -n demo apply -f redis-source.yaml
kubectl --context source -n demo wait redissentinel/redis-source-sentinel --for=jsonpath='{.status.phase}'=Ready --timeout=15m
kubectl --context source -n demo wait redis/redis-source --for=jsonpath='{.status.phase}'=Ready --timeout=30m

Seed and inspect the source data

Discover the auth Secret referenced by redis-source:

kubectl --context source -n demo get redis redis-source -o jsonpath='{.spec.authSecret.name}{"\n"}'
redis-source-auth

Use the returned name below if it differs. Read its password:

kubectl --context source -n demo get secret redis-source-auth -o jsonpath='{.data.password}' | base64 -d

Enter this password when the client prompts.

From your terminal, open a pod shell:

kubectl --context source -n demo exec -it redis-source-0 -c redis -- bash

Inside the pod, start the client through the writable database Service:

redis-cli -h redis-source.demo.svc -p 6379 --askpass

Enter Redis commands at its prompt. Connections through the source Service reach its writable primary even when the pod you opened is a replica.

Create five customers. Expected results below omit connection banners and are formatted for readability.

HSET customer:1 full_name "Customer 1" email "[email protected]"
HSET customer:2 full_name "Customer 2" email "[email protected]"
HSET customer:3 full_name "Customer 3" email "[email protected]"
HSET customer:4 full_name "Customer 4" email "[email protected]"
HSET customer:5 full_name "Customer 5" email "[email protected]"
WAIT 2 5000

WAIT should return 2, confirming that both replicas acknowledged the writes from this client connection. If it returns less, wait for replication and repeat WAIT 2 5000 before continuing.

HMGET customer:1 full_name email
HMGET customer:2 full_name email
HMGET customer:3 full_name email
HMGET customer:4 full_name email
HMGET customer:5 full_name email
1) "Customer 1"
2) "[email protected]"
1) "Customer 2"
2) "[email protected]"
1) "Customer 3"
2) "[email protected]"
1) "Customer 4"
2) "[email protected]"
1) "Customer 5"
2) "[email protected]"

Run QUIT to return to the pod’s Bash shell; keep that shell open.

Persist the source member used for the snapshot

Courier snapshots the source’s ordinal-0 data volume for this Sentinel example. Confirm that replication has reached it and persist its expected data before creating the Branch or waiting for refresh.

In the source pod’s Bash shell, connect directly to its local member rather than through the primary Service:

redis-cli -h 127.0.0.1 -p 6379 --askpass

Enter the source password, then check all five customers. The values must match the source results above:

HMGET customer:1 full_name email
HMGET customer:2 full_name email
HMGET customer:3 full_name email
HMGET customer:4 full_name email
HMGET customer:5 full_name email
1) "Customer 1"
2) "[email protected]"
1) "Customer 2"
2) "[email protected]"
1) "Customer 3"
2) "[email protected]"
1) "Customer 4"
2) "[email protected]"
1) "Customer 5"
2) "[email protected]"

If those values are still stale, wait for replication and repeat the reads. Once they match, write this member’s RDB snapshot:

SAVE
OK

Run QUIT, then exit to return to your terminal. This member must remain synchronized until Courier takes its snapshot.

Create the cross-cluster Branch

Save the following as redis-cross.yaml:

apiVersion: courier.kubedb.com/v1alpha1
kind: Branch
metadata:
  name: redis-cross
  namespace: demo
spec:
  source:
    databaseRef:
      apiGroup: kubedb.com
      kind: Redis
      name: redis-source
    namespace: demo
  target:
    namespace: dev
    name: redis-cross
    storageClassName: ceph-rbd
    resources:
      requests:
        cpu: 500m
        memory: 1Gi
      limits:
        memory: 1Gi
    clusterName: target
  deletionPolicy: Delete
kubectl --context source -n demo apply -f redis-cross.yaml
kubectl --context source -n demo wait branch/redis-cross --for=condition=Ready --timeout=60m
kubectl --context source -n demo get branch redis-cross

The Branch is created on source. Courier sends the target resource manifests through the OCM hub; the target KubeDB provisioner creates the database using the delivered cloned PVCs. target.clusterName is the OCM ManagedCluster name.

kubectl --context hub -n source get branchworks
kubectl --context hub -n target get manifestworks
kubectl --context target -n dev get redis redis-cross

Verify the copied data

Discover the auth Secret referenced by redis-cross:

kubectl --context target -n dev get redis redis-cross -o jsonpath='{.spec.authSecret.name}{"\n"}'
redis-cross-auth

Use the returned name below if it differs. Read its password:

kubectl --context target -n dev get secret redis-cross-auth -o jsonpath='{.data.password}' | base64 -d

Enter this password when the client prompts.

From your terminal, open a pod shell:

kubectl --context target -n dev exec -it redis-cross-0 -c redis -- bash

Inside the pod, start the client through the writable database Service:

redis-cli -h redis-cross.dev.svc -p 6379 --askpass

Enter Redis commands at its prompt. Connections through the source Service reach its writable primary even when the pod you opened is a replica.

HMGET customer:1 full_name email
HMGET customer:2 full_name email
HMGET customer:3 full_name email
HMGET customer:4 full_name email
HMGET customer:5 full_name email
1) "Customer 1"
2) "[email protected]"
1) "Customer 2"
2) "[email protected]"
1) "Customer 3"
2) "[email protected]"
1) "Customer 4"
2) "[email protected]"
1) "Customer 5"
2) "[email protected]"

Run QUIT to leave redis-cli, then exit to leave the pod shell.

Inspect target resources

kubectl --context target -n dev get pods,pvc -l app.kubernetes.io/instance=redis-cross

Read and write checks below verify the target is usable independently of the source.

Verify independent writes

Change the branch

Discover the auth Secret referenced by redis-cross:

kubectl --context target -n dev get redis redis-cross -o jsonpath='{.spec.authSecret.name}{"\n"}'
redis-cross-auth

Use the returned name below if it differs. Read its password:

kubectl --context target -n dev get secret redis-cross-auth -o jsonpath='{.data.password}' | base64 -d

Enter this password when the client prompts.

From your terminal, open a pod shell:

kubectl --context target -n dev exec -it redis-cross-0 -c redis -- bash

Inside the pod, start the client through the writable database Service:

redis-cli -h redis-cross.dev.svc -p 6379 --askpass

Enter Redis commands at its prompt. Connections through the source Service reach its writable primary even when the pod you opened is a replica.

HSET customer:5 full_name "Branch-only Customer 5"
HMGET customer:1 full_name email
HMGET customer:2 full_name email
HMGET customer:3 full_name email
HMGET customer:4 full_name email
HMGET customer:5 full_name email
1) "Customer 1"
2) "[email protected]"
1) "Customer 2"
2) "[email protected]"
1) "Customer 3"
2) "[email protected]"
1) "Customer 4"
2) "[email protected]"
1) "Branch-only Customer 5"
2) "[email protected]"

Run QUIT to leave redis-cli, then exit to leave the pod shell.

Change and inspect the source

Discover the auth Secret referenced by redis-source:

kubectl --context source -n demo get redis redis-source -o jsonpath='{.spec.authSecret.name}{"\n"}'
redis-source-auth

Use the returned name below if it differs. Read its password:

kubectl --context source -n demo get secret redis-source-auth -o jsonpath='{.data.password}' | base64 -d

Enter this password when the client prompts.

From your terminal, open a pod shell:

kubectl --context source -n demo exec -it redis-source-0 -c redis -- bash

Inside the pod, start the client through the writable database Service:

redis-cli -h redis-source.demo.svc -p 6379 --askpass

Enter Redis commands at its prompt. Connections through the source Service reach its writable primary even when the pod you opened is a replica.

HSET customer:1 full_name "Updated Customer 1"
WAIT 2 5000
HMGET customer:1 full_name email
HMGET customer:2 full_name email
HMGET customer:3 full_name email
HMGET customer:4 full_name email
HMGET customer:5 full_name email
1) "Updated Customer 1"
2) "[email protected]"
1) "Customer 2"
2) "[email protected]"
1) "Customer 3"
2) "[email protected]"
1) "Customer 4"
2) "[email protected]"
1) "Customer 5"
2) "[email protected]"

Confirm WAIT returned 2. Run QUIT to return to the pod’s Bash shell.

Persist the source member used for the snapshot

Courier snapshots the source’s ordinal-0 data volume for this Sentinel example. Confirm that replication has reached it and persist its expected data before creating the Branch or waiting for refresh.

In the source pod’s Bash shell, connect directly to its local member rather than through the primary Service:

redis-cli -h 127.0.0.1 -p 6379 --askpass

Enter the source password, then check all five customers. The values must match the source results above:

HMGET customer:1 full_name email
HMGET customer:2 full_name email
HMGET customer:3 full_name email
HMGET customer:4 full_name email
HMGET customer:5 full_name email
1) "Updated Customer 1"
2) "[email protected]"
1) "Customer 2"
2) "[email protected]"
1) "Customer 3"
2) "[email protected]"
1) "Customer 4"
2) "[email protected]"
1) "Customer 5"
2) "[email protected]"

If those values are still stale, wait for replication and repeat the reads. Once they match, write this member’s RDB snapshot:

SAVE
OK

Run QUIT, then exit to return to your terminal. This member must remain synchronized until Courier takes its snapshot.

The source retains the original customer 5; the branch edit did not reach it.

Check the branch again

Discover the auth Secret referenced by redis-cross:

kubectl --context target -n dev get redis redis-cross -o jsonpath='{.spec.authSecret.name}{"\n"}'
redis-cross-auth

Use the returned name below if it differs. Read its password:

kubectl --context target -n dev get secret redis-cross-auth -o jsonpath='{.data.password}' | base64 -d

Enter this password when the client prompts.

From your terminal, open a pod shell:

kubectl --context target -n dev exec -it redis-cross-0 -c redis -- bash

Inside the pod, start the client through the writable database Service:

redis-cli -h redis-cross.dev.svc -p 6379 --askpass

Enter Redis commands at its prompt. Connections through the source Service reach its writable primary even when the pod you opened is a replica.

HMGET customer:1 full_name email
HMGET customer:2 full_name email
HMGET customer:3 full_name email
HMGET customer:4 full_name email
HMGET customer:5 full_name email
1) "Customer 1"
2) "[email protected]"
1) "Customer 2"
2) "[email protected]"
1) "Customer 3"
2) "[email protected]"
1) "Customer 4"
2) "[email protected]"
1) "Branch-only Customer 5"
2) "[email protected]"

Run QUIT to leave redis-cli, then exit to leave the pod shell.

The branch retains its own customer 5 and the original customer 1. The source update did not reach it.

Cleanup

Delete the Branch on the source cluster first. Let Courier remove its target and snapshots before deleting the source:

kubectl --context source -n demo delete branch redis-cross --wait=true --timeout=30m
kubectl --context source -n demo delete redis redis-source --wait=true --timeout=30m
kubectl --context source -n demo delete redissentinel redis-source-sentinel --wait=true --timeout=15m

Keep Courier, CSI, the shared backend, and OCM when used available until cleanup completes. These commands leave the namespaces in place.

Next Steps