Skip to main content

Configuration: Cluster

  • Avoid using different cluster configurations across nodes:
    Configuration mismatches can lead to interaction problems between nodes.

  • If you must set different configurations for individual nodes, we recommend testing the setup in a development environment first to ensure proper interaction between all nodes.

Cluster.CompareExchangeExpiredDeleteFrequencyInSec​

Time (in seconds) between cleanup of expired compare exchange items.

  • Type: int
  • Default: 60
  • Scope: Server-wide only

Cluster.CompareExchangeTombstonesCleanupIntervalInMin​

Time (in minutes) between cleanup of compare exchange tombstones.

  • Type: int
  • Default: 10
  • Scope: Server-wide only

Cluster.ElectionTimeoutInMs​

Timeout (in milliseconds) within which the node expects to receive a heartbeat from the leader.

  • Type: int
  • Default: 300
  • Scope: Server-wide only

Cluster.HardDeleteOnReplacement​

Set hard/soft delete for a database that was removed by the observer from the cluster topology in order to maintain the replication factor.

  • Type: bool
  • Default: true
  • Scope: Server-wide only

Cluster.LogHistoryMaxEntries​

Maximum number of log entries to keep in the history log table.

  • Type: int
  • Default: 2048
  • Scope: Server-wide only

Cluster.MaxChangeVectorDistance​

Exceeding the allowed change vector distance between two nodes will move the lagged node to rehab.

  • Type: long
  • Default: 65536
  • Scope: Server-wide only

Cluster.MaxClusterTransactionCompareExchangeTombstoneCheckIntervalInMin​

The maximum interval (in minutes) between checks for compare exchange tombstones that are performed by the cluster-wide transaction mechanism.

  • Type: int
  • Default: 5
  • Scope: Server-wide only

Cluster.MaxSizeOfSingleRaftCommandInMb​

EXPERT ONLY:
The maximum allowed size (in megabytes) for a single raft command.

  • Type: int
  • SizeUnit(SizeUnit.Megabytes)
  • Default: 128
  • Scope: Server-wide only

Cluster.MaximalAllowedClusterVersion​

EXPERT ONLY:
If exceeded, restrict the cluster to the specified version.

  • Type: int?
  • Default: null
  • Scope: Server-wide only

Cluster.OnErrorDelayTimeInMs​

How long the maintenance supervisor waits (in milliseconds) after receiving an exception from a worker before retrying.

  • Type: int
  • Default: 5000
  • Scope: Server-wide only

Cluster.OperationTimeoutInSec​

As a cluster node, how long (in seconds) to wait before timing out an operation between two cluster nodes.

  • Type: int
  • Default: 15
  • Scope: Server-wide only

Cluster.ReceiveFromWorkerTimeoutInMs​

How long the maintenance supervisor waits (in milliseconds) for a response from a worker before timing out.

  • Type: int
  • Default: 5000
  • Scope: Server-wide only

Cluster.StatsStabilizationTimeInSec​

How long to wait (in seconds) for cluster stats to stabilize after a database topology change.

  • Type: int
  • Default: 5
  • Scope: Server-wide only

Cluster.SupervisorSamplePeriodInMs​

How long the maintenance supervisor waits (in milliseconds) between sampling the information received from the nodes.

  • Type: int
  • Default: 1000
  • Scope: Server-wide only

Cluster.TcpReceiveBufferSizeInBytes​

The size (in bytes) of the TCP connection receive buffer.

  • Type: int
  • Default: 32 * 1024
  • Scope: Server-wide only

Cluster.TcpSendBufferSizeInBytes​

The size (in bytes) of the TCP connection send buffer.

  • Type: int
  • Default: 32 * 1024
  • Scope: Server-wide only

Cluster.TcpTimeoutInMs​

TCP connection read/write timeout (in milliseconds).

  • Type: int
  • Default: 15_000
  • Scope: Server-wide only

Cluster.TimeBeforeAddingReplicaInSec​

The time (in seconds) a database instance must be in a good and responsive state before we add a replica to match the replication factor.

  • Type: int
  • Default: 900
  • Scope: Server-wide only

Cluster.TimeBeforeMovingToRehabInSec​

The grace period (in seconds) we give a node before it is moved to rehab.

  • Type: int
  • Default: 60
  • Scope: Server-wide only

Cluster.TimeBeforeRotatingPreferredNodeInSec​

The grace period (in seconds) we give the preferred node before moving it to the end of the members list.

  • Type: int
  • Default: 5
  • Scope: Server-wide only

Cluster.WorkerSamplePeriodInMs​

The time (in milliseconds) between sampling database information and sending it to the maintenance supervisor.

  • Type: int
  • Default: 500
  • Scope: Server-wide only

Cluster.DisableAtomicDocumentWrites​

EXPERT ONLY:
Disable automatic atomic writes with cluster write transactions.
If set to true, will only consider explicitly added compare exchange values to validate cluster wide transactions.

  • Type: bool
  • Default: false
  • Scope: Server-wide or per database

Cluster.MaxClusterTransactionBatchSize​

EXPERT ONLY:
Specifies the maximum size of the cluster transaction batch to be executed on the database at once.

  • Type: int
  • Default: 256
  • Scope: Server-wide or per database