Kvindo™ Cloud Docs
The same deployment through the Python SDK: client.valkey_replications — a separate accessor from client.valkeys, because this is a different product with its own permission enum and its own microservice. It shares only valkey_parameters_sets and valkey_users with the cluster tier. Assumes the Networking example's client/folder_id/subnet_id are already in scope. Unlike the CLI, the SDK takes real ids everywhere — instances[].vpcSubnetId is just subnet_id, no lookup needed.
Once reconciled, read the connection URI a client would actually use: client.valkey_replications.read(repl_id).status["connectionUri"]. It is a snapshot — after a Sentinel failover it lags the real topology, so a client that treats it as a stable write address will eventually write to a demoted node and get READONLY.
A manual failover is declarative: set requestedPrimaryInstanceId to the instance you want promoted, and the platform issues a server-side FAILOVER TO. It is consumed once and then cleared. There is no automatic failback — a failback is a second failover.
Resource docs
More in this section
Reference