Important
The SingleStore 9.1 release candidate (RC) gives you the opportunity to preview, evaluate, and provide feedback on new and upcoming features prior to their general availability. In the interim, SingleStore 9.0 is recommended for production workloads, which can later be upgraded to SingleStore 9.1.
Best Practices for Successful Rebalance Operations
-
Upgrade Sequence: Perform the engine upgrade to at least version 9.
0 before initiating scale-out activities; versions 8. 9 and 9. 0 contain critical fixes that ensure a smoother scaling process. -
Redundancy Level: Increase the cluster configuration to Redundancy 2.
The setting of Redundancy 1 is not suitable for large production environments and poses significant risk during upgrades or failovers. -
Storage Architecture: Utilize the standard configuration with local disks or high-quality direct-attached SAN LUNs/JBODs.
SingleStore recommends against using "bottomless" storage or remote Linux file systems for on-premise deployments. -
Memory Configuration:
-
Set
maximum_to 90% of the host's physical RAM.memory -
Set
maximum_to 90% of the configuredtable_ memory maximum_.memory
-
-
Resource Management: Configure Resource Pools with strict memory and CPU restrictions to isolate workloads and protect cluster stability.
-
Partition Sizing: Adhere to best practices regarding the number of partitions per core.
Utilizing smaller, more numerous partitions reduces rebalance times and lowers the risk of failure during partition moves. -
Operational Maintenance: Use the
REBALANCE PARTITIONS .command during maintenance; this ensures slow mergers are terminated automatically to prevent blocking.. . WITH KILL -
Professional Services: SingleStore recommends engaging Support for major activities (such as upgrades or expansions) to ensure continuous, dedicated on-call support.
Last modified: March 13, 2026