Precautions

This function is in the open beta test (OBT) phase. To use it, contact customer service to apply for the required permissions.
Before you create a dual-active DR relationship, prepare two GeminiDB Redis instances in two different VPCs and make sure that their IP addresses do not start with 192 or 172 and their CIDR blocks do not overlap. If the two VPCs are in the same region, create a peering connection between them so that they can be connected. If they are not, create a cross-region cloud connection between them. For details, see Network Configurations.
Ensure that the two prepared instances have the same specifications, storage space, and nodes and both use the database port 8635.
When a dual-active DR relationship is being created, the primary instance may be interrupted for seconds multiple times. This requires that your services can retry automatically. The standby instance cannot handle service requests, and data on it will be deleted and finally kept consistent with that on the primary instance. The time required for creating a dual-active DR relationship depends on the nodes and data volume of the two instances.
If a dual-active relationship has been created for two instances, do not change their specifications or add or delete nodes for them.
After a dual-active DR relationship is created, you can scale storage space of the primary and standby instances. Make sure that the two instances have the same storage space in the long run.
After a dual-active DR relationship is created, running command flushall is not supported.
If both the primary and standby instances support writes, do not modify the same key at the same time or near the same time. Otherwise, data on the two instances may become inconsistent. Running command flushdb on the standby instance is not supported.
If the primary and standby instances are not in the same regions, the synchronization latency depends on the cloud connection latency and on whether write traffic matches the bandwidth configured for the cloud connection. If the write traffic is beyond the bandwidth, synchronization data will be stacked. You can view metric Size of WAL Files to Be Synchronized and set alarm rules. For details, see Viewing Monitoring Metrics.
HyperMetro supports only cluster instances.