Class ConsistentHashingShardAllocationStrategy

java.lang.Object
org.apache.pekko.cluster.sharding.ConsistentHashingShardAllocationStrategy
All Implemented Interfaces:
NoSerializationVerificationNeeded, ClusterShardAllocationMixin, ShardCoordinator.ActorSystemDependentAllocationStrategy, ShardCoordinator.ShardAllocationStrategy

public class ConsistentHashingShardAllocationStrategy extends Object implements ShardCoordinator.ActorSystemDependentAllocationStrategy, ClusterShardAllocationMixin
pekko.cluster.sharding.ShardCoordinator.ShardAllocationStrategy that is using consistent hashing. This can be useful when shards with the same shard id for different entity types should be best effort colocated to the same nodes.

When adding or removing nodes it will rebalance according to the new consistent hashing, but that means that only a few shards will be rebalanced and others remain on the same location.

A good explanation of Consistent Hashing: https://tom-e-white.com/2007/11/consistent-hashing.html

Create a new instance of this for each entity types, i.e. a ConsistentHashingShardAllocationStrategy instance must not be shared between different entity types.

Not intended for public inheritance/implementation

  • Constructor Details

    • ConsistentHashingShardAllocationStrategy

      public ConsistentHashingShardAllocationStrategy(int rebalanceLimit)
  • Method Details

    • start

      public void start(ActorSystem system)
      Called before any calls to allocate/rebalance. Do not block. If asynchronous actions are required they can be started here and delay the Futures returned by allocate/rebalance.
      Specified by:
      start in interface ShardCoordinator.ActorSystemDependentAllocationStrategy
    • log

      protected LoggingAdapter log()
    • clusterState

      protected ClusterEvent.CurrentClusterState clusterState()
      Specified by:
      clusterState in interface ClusterShardAllocationMixin
    • selfMember

      protected Member selfMember()
      Specified by:
      selfMember in interface ClusterShardAllocationMixin
    • allocateShard

      public scala.concurrent.Future<ActorRef> allocateShard(ActorRef requester, String shardId, scala.collection.immutable.Map<ActorRef,scala.collection.immutable.IndexedSeq<String>> currentShardAllocations)
      Description copied from interface: ShardCoordinator.ShardAllocationStrategy
      Invoked when the location of a new shard is to be decided.

      Specified by:
      allocateShard in interface ShardCoordinator.ShardAllocationStrategy
      Parameters:
      requester - actor reference to the ShardRegion that requested the location of the shard, can be returned if preference should be given to the node where the shard was first accessed
      shardId - the id of the shard to allocate
      currentShardAllocations - all actor refs to ShardRegion and their current allocated shards, in the order they were allocated
      Returns:
      a Future of the actor ref of the ShardRegion that is to be responsible for the shard, must be one of the references included in the currentShardAllocations parameter
    • rebalance

      public scala.concurrent.Future<scala.collection.immutable.Set<String>> rebalance(scala.collection.immutable.Map<ActorRef,scala.collection.immutable.IndexedSeq<String>> currentShardAllocations, scala.collection.immutable.Set<String> rebalanceInProgress)
      Description copied from interface: ShardCoordinator.ShardAllocationStrategy
      Invoked periodically to decide which shards to rebalance to another location.

      Specified by:
      rebalance in interface ShardCoordinator.ShardAllocationStrategy
      Parameters:
      currentShardAllocations - all actor refs to ShardRegion and their current allocated shards, in the order they were allocated
      rebalanceInProgress - set of shards that are currently being rebalanced, i.e. you should not include these in the returned set
      Returns:
      a Future of the shards to be migrated, may be empty to skip rebalance in this round