Auth Migration

Legacy AUTH token authentication uses a single shared password for all clients. RBAC provides per-user access control with fine-grained command and key permissions. This migration moves from the legacy model to the recommended model.

Prerequisites

• Node-based replication group with TransitEncryptionEnabled: true
• Engine version Redis OSS 6.0 or higher, or Valkey 7.2 or higher (RBAC is not supported on earlier versions)
• AUTH token currently configured on the replication group
• Application code ready to switch to username/password authentication

Step 1: Create RBAC Users

Create one or more RBAC users that match your application’s access needs.

For IAM auth instead of passwords:

Store RBAC passwords in Secrets Manager:

Step 2: Create User Group and Associate

Note: At this point, both AUTH token and RBAC credentials are active. Existing clients using the AUTH token continue to work.

Step 3: Update Application to Use RBAC Credentials

Update application code to pass username and password (or IAM token) instead of just the AUTH token.

Before (AUTH token):

After (RBAC):

Deploy the updated application. Verify that it connects successfully using RBAC credentials.

Step 4: Verify RBAC Connectivity

Before removing the AUTH token, verify that all clients are successfully authenticating via RBAC:

Confirm that the application works correctly with RBAC credentials and that no clients still depend on the AUTH token.

Step 5: Remove AUTH Token

Now that all clients are using RBAC, remove the AUTH token:

Step 6: Verify AUTH Token Removal

Verify that the AUTH token is disabled:

Step 7: Verify and Secure the Default User

The default user is always present in RBAC user groups. Disable it or restrict its access to prevent unauthenticated connections:

For Valkey clusters: The default user can be removed from the user group entirely, or disabled:

For Redis OSS clusters: The default user must remain in the user group (Redis OSS requires it), but should be disabled:

Note: On Redis OSS, removing the default user from a user group will fail. On Valkey, the default user is optional in user groups, so removing it is the cleanest approach.

Rollback Plan

If issues arise during migration:

1. Before Step 5: The AUTH token is still active. Simply revert application code to use AUTH token authentication and deploy. Both auth methods work simultaneously, so rollback is safe at any point before Step 5

2. After Step 5: The AUTH token has been removed. To roll back, re-set the AUTH token on the replication group, revert application code, and deploy

3. Remove the user group association if needed:

   bashCopy

   aws elasticache modify-replication-group \
     --replication-group-id <cluster-id> \
     --user-group-ids-to-remove myapp-usergroup \
     --region <region>

Cluster-Mode Disabled to Cluster-Mode Enabled Migration

Cluster-mode disabled uses a single shard (primary + replicas). Cluster-mode enabled uses multiple shards for horizontal scaling. This migration changes the endpoint model and requires client updates.

Prerequisites

• Understand that cluster-mode enabled changes the client interaction model
• Client libraries must support MOVED and ASK redirect handling
• Multi-key commands must use hash tags to colocate keys on the same shard
• For in-place migration: Minimum engine version of Valkey 7.2 or Redis OSS 7.0 is required
• For in-place migration: Auto-failover must be enabled with at least 1 replica
• For in-place migration: The cluster may only have keys in database 0 (multiple databases are not supported)
• For in-place migration: CMD→CME is a one-way operation. Once cluster mode is set to enabled, it cannot be reverted back to disabled. You can only revert from compatible back to disabled

Impact Assessment

Step 1: Assess Client Compatibility

Verify that all client applications can handle cluster mode:

Check for multi-key commands that span keys without hash tags:

• MGET key1 key2 – keys must be on the same shard
• SUNION set1 set2 – sets must be on the same shard
• Lua scripts accessing multiple keys – all keys must be in the same slot
• MULTI/EXEC transactions – all keys must be in the same slot

Fix: use hash tags like {user:100}:profile and {user:100}:sessions to ensure related keys land on the same shard.

Step 2: Create a New Cluster-Mode Enabled Cluster

For Valkey 7.2+ or Redis OSS 7.0+, you can perform an in-place migration from cluster-mode disabled to cluster-mode enabled using the modify-replication-group CLI command (or ModifyReplicationGroup API). This is a two-step process: first set cluster mode to compatible, then to enabled. Requirements: auto-failover must be enabled with at least 1 replica, keys must exist only in database 0, and the migration is irreversible once set to enabled (you can revert from compatible back to disabled).

Note: In compatible mode, other modification operations such as scaling and engine version upgrades are not allowed. The cluster endpoints will change once cluster mode is set to enabled. Make sure to update your applications with the new endpoints.

For older engine versions (below Redis OSS 7.0), you must create a new cluster and migrate data.

Step 3: Migrate Data

Option A: Snapshot and restore (brief downtime acceptable)

Option B: Dual-write migration (zero downtime)

1. Configure the application to write to both old and new clusters
2. Run a backfill to copy existing data from old to new (application-level copy)
3. Gradually shift reads to the new cluster
4. Stop writes to the old cluster
5. Decommission the old cluster

Warning: Do NOT use DUMP/RESTORE for data migration between clusters. The serialization format is engine-version-specific and may produce corrupted data on version mismatches. Use application-level copy (read from source, write to target) or the ElastiCache online migration tools instead. Note: The ElastiCache online migration tool (start-migration / complete-migration) is designed exclusively for migrating data from self-hosted open-source Valkey or Redis OSS on Amazon EC2 to ElastiCache. It is not for moving data between ElastiCache clusters. Additionally, the target cluster must not have encryption in-transit enabled, must have Multi-AZ enabled, must not be part of a global datastore, must have data tiering disabled, and the number of shards in source and target must match. Online migration is not supported for ElastiCache serverless caches or clusters running on the r6gd node type. See the Online Migration documentation (opens in a new tab) for details.

Option C: Online migration via ElastiCache (only if source is self-hosted Valkey/Redis on EC2, not an ElastiCache cluster)

Step 4: Update Application Endpoints

The new cluster uses a configuration endpoint that resolves to all shards. Update connection strings:

Before:

After:

Step 5: Validate Application Behavior

• Verify all read and write operations succeed
• Check for CROSSSLOT errors (indicates multi-key operations on different shards)
• Monitor CloudWatch metrics for error rates, latency, and connection counts
• Run valkey-cli -h <new-endpoint> -p 6379 --tls PING for basic validation

Rollback Plan

For the in-place migration path (Valkey 7.2+/Redis OSS 7.0+): Once cluster mode is set to enabled, the change is irreversible — you cannot convert back to cluster-mode disabled. You can only revert from compatible back to disabled before completing the final step. Take a snapshot before setting enabled.

For the new-cluster migration path: The old cluster remains running throughout the migration:

1. Revert application endpoints to the old cluster
2. Deploy the reverted application
3. Delete the new cluster-mode enabled cluster
4. No data loss (old cluster was never modified)

Note on Endpoint Model Change

This migration changes the endpoint model:

• Cluster-mode disabled: single primary endpoint + single reader endpoint
• Cluster-mode enabled: configuration endpoint that the client uses to discover all shards

Applications must update their connection logic, not just the endpoint hostname. This is the most common source of issues during this migration.