A Tair (Redis OSS-compatible) instance moves through a set of well-defined states as you create, configure, and manage it. Understanding these states helps you anticipate service behavior and take the right action when a state change affects your application.
Check the current state
Two methods are available:
Console: Open the Instances page and read the state label next to the instance.
API: Call DescribeInstances and read the
Statusfield in the response.
Stable states
Stable states represent a resting condition — no operation is in progress.
| Console state | API value | Description |
|---|---|---|
| Creating | Creating | The instance is being provisioned. Database services are unavailable until the instance reaches Running. |
| Running | Normal | The instance is healthy and serving requests. |
| Disabled | Inactive | The instance has expired and is no longer serving requests. Renew the instance to restore service. For details, see Renewal. |
| Released | N/A | The instance has been released (manually or automatically when overdue or expired) and is no longer visible on the Instances page. To restore it, find and recover the instance from the recycle bin. For details, see Manage instances in the recycle bin. |
Transitional states
Transitional states are temporary. The instance is undergoing an operation and returns to Running when the operation completes.
The Flushing Instance and Deleting Expired Data states involve permanent data deletion. Deleted data cannot be restored. Perform these operations during off-peak hours.
| Console state | API value | Triggered by | Impact |
|---|---|---|---|
| Converting | Transforming | Change the billing method to subscription or pay-as-you-go | No impact on service availability. |
| Changing Configuration | Changing | Change the configurations of an instance or adjust the number of shards | Latency may be unstable during the operation. Scaling does not cause transient connections or put the instance in a read-only state. Perform this operation during off-peak hours and make sure your application reconnects automatically. |
| Restarting | Rebooting | Restart an instance | Transient connections last a few seconds. Perform this operation during off-peak hours and make sure your application reconnects automatically. |
| Upgrading Major Version | MajorVersionUpgrading | Upgrade the major version | See the major version upgrade documentation for specific impacts. |
| Updating Minor Version | MinorVersionUpgrading | Update the minor version | |
| Configuring Network | NetworkModifying | Multiple network operations (see Network operation impacts) | Varies by operation. |
| Configuring SSL | SSLModifying | Enable TLS encryption | The instance restarts after you enable Transport Layer Security (TLS) encryption or update the certificate validity period. Transient connections last a few seconds. Perform this operation during off-peak hours and make sure your application reconnects automatically. |
| Modifying TDE | TEDModifying | Enable TDE | No impact on service availability. |
| Migrating to Another Zone | ZoneMigrating | Migrate an instance across zones | See Zone migration impacts. |
| Flushing Instance | Flushing | Clear all data | The instance runs FLUSHALL to delete all data immediately. Deleted data cannot be restored. This operation affects online services. Perform it during off-peak hours. |
| Deleting Expired Data | CleaningUpExpiredData | Clear expired data | The instance runs SCAN to delete all expired data immediately. Deleted data cannot be restored. This operation affects online services. Perform it during off-peak hours. |
| Switching | HASwitching | Manually switch workloads from a master node to a replica node | See HA switchover impacts. |
Network operation impacts
When the instance is in the Configuring Network state, the specific impact depends on which network operation triggered it.
| Operation | Impact |
|---|---|
| Switch from classic network to VPC | The switch cannot be reversed. Transient connections last a few seconds. Decide whether to retain the classic network endpoint before switching — if you do not retain it, the endpoint is released after the switch, and clients using that endpoint lose connectivity. Update the database endpoint on all clients as soon as the switch completes. Perform this operation during off-peak hours and make sure your application reconnects automatically. |
| Change the endpoint or port number | Clients can no longer connect using the original endpoint or port. Update the connection information on all clients as soon as the change completes. |
| Apply for a public endpoint | No impact on existing connections. |
| Release a public endpoint | Clients can no longer connect using the original public endpoint. Update the connection information on all clients as soon as the release completes. |
| Enable direct connection mode | After enabling direct connection mode, you cannot upgrade the major version or change the zone of this instance. To perform either of those operations, release the private endpoint first. For details, see Upgrade the major version, Migrate an instance across zones, and Release a private endpoint. |
| Release a private endpoint | Clients can no longer connect using the original private endpoint. Update the connection information on all clients as soon as the release completes. |
| Change the VPC or vSwitch | Transient connections last a few seconds. Perform this operation during off-peak hours and make sure your application reconnects automatically. |
Zone migration impacts
When the instance is in the Migrating to Another Zone state:
Transient connections last a few seconds. Perform this operation during off-peak hours and make sure your application reconnects automatically.
The virtual IP address (VIP) of the instance changes (for example, from
172.16.88.60to a new address), but the endpoint remains unchanged. Always connect to the instance using the endpoint, not the VIP. Connections using the old VIP will fail after migration.If the instance's minor version is outdated, the system updates it to the latest version to maintain performance and stability.
HA switchover impacts
When the instance is in the Switching state:
Data nodes are disconnected for a few seconds during the switchover.
Data loss is possible. Synchronization latency between the master node and replica node may cause data inconsistency at the time of switchover. To reduce the risk of data loss and prevent data doublewrite caused by Domain Name System (DNS) cache, data nodes remain read-only for up to 30 seconds after the switchover.
All instance management operations are blocked while the instance is in the Switching state. This includes modifying configurations and migrating to another zone.