redis.clients.jedis.providers

Class MultiClusterPooledConnectionProvider

java.lang.Object

redis.clients.jedis.providers.MultiClusterPooledConnectionProvider

All Implemented Interfaces:

java.lang.AutoCloseable, ConnectionProvider

@Experimental
public class MultiClusterPooledConnectionProvider
extends java.lang.Object
implements ConnectionProvider

Author:

Allen Terleto (aterleto)

ConnectionProvider which supports multiple cluster/database endpoints each with their own isolated connection pool. With this ConnectionProvider users can seamlessly failover to Disaster Recovery (DR), Backup, and Active-Active cluster(s) by using simple configuration which is passed through from Resilience4j - https://resilience4j.readme.io/docs

Support for manual failback is provided by way of setActiveMultiClusterIndex(int)

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	MultiClusterPooledConnectionProvider.Cluster

Constructor Summary

Constructors

Constructor and Description
MultiClusterPooledConnectionProvider(MultiClusterClientConfig multiClusterClientConfig)

Method Summary

Modifier and Type	Method and Description
void	close()
MultiClusterPooledConnectionProvider.Cluster	getCluster()
MultiClusterPooledConnectionProvider.Cluster	getCluster(int multiClusterIndex)
io.github.resilience4j.circuitbreaker.CircuitBreaker	getClusterCircuitBreaker()
io.github.resilience4j.circuitbreaker.CircuitBreaker	getClusterCircuitBreaker(int multiClusterIndex)
Connection	getConnection()
Connection	getConnection(CommandArguments args)
Connection	getConnection(int multiClusterIndex)
java.util.Map<?,Pool<Connection>>	getConnectionMap()
java.util.List<java.lang.Class<? extends java.lang.Throwable>>	getFallbackExceptionList()
int	incrementActiveMultiClusterIndex() Increments the actively used cluster/database endpoint (connection pool) amongst the pre-configured list which were provided at startup via the MultiClusterClientConfig.
boolean	isLastClusterCircuitBreakerForcedOpen()
void	runClusterFailoverPostProcessor(java.lang.Integer multiClusterIndex)
void	setActiveMultiClusterIndex(int multiClusterIndex) Manually overrides the actively used cluster/database endpoint (connection pool) amongst the pre-configured list which were provided at startup via the MultiClusterClientConfig.
void	setClusterFailoverPostProcessor(java.util.function.Consumer<java.lang.String> clusterFailoverPostProcessor)
void	validateTargetConnection(int multiClusterIndex) Design decision was made to defer responsibility for cross-replication validation to the user.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

MultiClusterPooledConnectionProvider

public MultiClusterPooledConnectionProvider(MultiClusterClientConfig multiClusterClientConfig)

Method Detail

incrementActiveMultiClusterIndex

public int incrementActiveMultiClusterIndex()

Increments the actively used cluster/database endpoint (connection pool) amongst the pre-configured list which were provided at startup via the MultiClusterClientConfig. All traffic will be routed according to this index. Only indexes within the pre-configured range (static) are supported otherwise an exception will be thrown. In the event that the next prioritized connection has a forced open state, the method will recursively increment the index in order to avoid a failed command.

validateTargetConnection

public void validateTargetConnection(int multiClusterIndex)

Design decision was made to defer responsibility for cross-replication validation to the user. Alternatively there was discussion to handle cross-cluster replication validation by setting a key/value pair per hashslot in the active connection (with a TTL) and subsequently reading it from the target connection.

setActiveMultiClusterIndex

public void setActiveMultiClusterIndex(int multiClusterIndex)

Manually overrides the actively used cluster/database endpoint (connection pool) amongst the pre-configured list which were provided at startup via the MultiClusterClientConfig. All traffic will be routed according to the provided new index. Special care should be taken to confirm cluster/database availability AND potentially cross-cluster replication BEFORE using this capability.

close

public void close()

Specified by:

close in interface java.lang.AutoCloseable

getConnection

public Connection getConnection()

Specified by:

getConnection in interface ConnectionProvider

getConnection

public Connection getConnection(int multiClusterIndex)

getConnection

public Connection getConnection(CommandArguments args)

Specified by:

getConnection in interface ConnectionProvider

getConnectionMap

public java.util.Map<?,Pool<Connection>> getConnectionMap()

Specified by:

getConnectionMap in interface ConnectionProvider

getCluster

public MultiClusterPooledConnectionProvider.Cluster getCluster()

getCluster

@VisibleForTesting
public MultiClusterPooledConnectionProvider.Cluster getCluster(int multiClusterIndex)

getClusterCircuitBreaker

public io.github.resilience4j.circuitbreaker.CircuitBreaker getClusterCircuitBreaker()

getClusterCircuitBreaker

public io.github.resilience4j.circuitbreaker.CircuitBreaker getClusterCircuitBreaker(int multiClusterIndex)

isLastClusterCircuitBreakerForcedOpen

public boolean isLastClusterCircuitBreakerForcedOpen()

runClusterFailoverPostProcessor

public void runClusterFailoverPostProcessor(java.lang.Integer multiClusterIndex)

setClusterFailoverPostProcessor

public void setClusterFailoverPostProcessor(java.util.function.Consumer<java.lang.String> clusterFailoverPostProcessor)

getFallbackExceptionList

public java.util.List<java.lang.Class<? extends java.lang.Throwable>> getFallbackExceptionList()