Package com.clickhouse.client

Class ClickHouseLoadBalancingPolicy

java.lang.Object

com.clickhouse.client.ClickHouseLoadBalancingPolicy

All Implemented Interfaces:

Serializable

public abstract class ClickHouseLoadBalancingPolicy
extends Object
implements Serializable

Load balancing policy. In general, a policy is responsible for 3 things: 1) get node from a managed list; 2) managing node's status; and 3) optionally schedule background tasks like node discovery and health check.

See Also:

Serialized Form

Field Summary

Fields

Modifier and Type	Field	Description
static String	FIRST_ALIVE	Similar as the default policy, which always picks the first healthy node.
static String	QUERY_GET_ALL_NODES
static String	QUERY_GET_CLUSTER_NODES
static String	QUERY_GET_OTHER_NODES
static String	RANDOM	Policy to pick a healthy node randomly from the list.
static String	ROUND_ROBIN	Policy to pick healthy node one after another based their order in the list.

Constructor Summary

Constructors

Constructor	Description
ClickHouseLoadBalancingPolicy()

Method Summary

Modifier and Type	Method	Description
protected ClickHouseNode	get(ClickHouseNodes manager)	Gets next node available in the list.
protected ClickHouseNode	get(ClickHouseNodes manager, ClickHouseNodeSelector t)	Gets next node available in the list according to the given node selector.
protected String	getQueryForAllLocalNodes()	Gets a SQL query for finding all local nodes regardless which cluster it belongs to.
protected String	getQueryForClusterLocalNodes()	Gets a parameterized SQL query for finding all local nodes of a specific cluster.
protected String	getQueryForNonLocalNodes()	Gets a parameterized SQL query for finding all non-local nodes in one or many clusters.
protected ScheduledExecutorService	getScheduler()	Gets scheduled executor service for auto discovery and health check.
static ClickHouseLoadBalancingPolicy	of(String name)	Gets or creates singleton load balancing policy.
protected ScheduledFuture<?>	schedule(ScheduledFuture<?> current, Runnable task, long interval)	Schedules the task to run only when current run has been completed/cancelled or does not exist.
protected ClickHouseNode	suggest(ClickHouseNodes manager, ClickHouseNode server, Throwable failure)	Sugguests a new node as replacement of the given one in order to recover from the failure, which is usually a connection error.
String	toString()
protected void	update(ClickHouseNodes manager, ClickHouseNode node, ClickHouseNode.Status status)	Updates node status to one of ClickHouseNode.Status.

Methods inherited from class java.lang.Object

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

Field Detail

QUERY_GET_ALL_NODES

public static final String QUERY_GET_ALL_NODES

See Also:

Constant Field Values

QUERY_GET_CLUSTER_NODES

public static final String QUERY_GET_CLUSTER_NODES

See Also:

Constant Field Values

QUERY_GET_OTHER_NODES

public static final String QUERY_GET_OTHER_NODES

See Also:

Constant Field Values

FIRST_ALIVE

public static final String FIRST_ALIVE

Similar as the default policy, which always picks the first healthy node. Besides, it provides scheduler for node discovery and health check.

See Also:

Constant Field Values

RANDOM

public static final String RANDOM

Policy to pick a healthy node randomly from the list.

See Also:

Constant Field Values

ROUND_ROBIN

public static final String ROUND_ROBIN

Policy to pick healthy node one after another based their order in the list.

See Also:

Constant Field Values

Constructor Detail

ClickHouseLoadBalancingPolicy

public ClickHouseLoadBalancingPolicy()

Method Detail

of

public static ClickHouseLoadBalancingPolicy of(String name)

Gets or creates singleton load balancing policy.

Parameters:

name - policy name, one of FIRST_ALIVE,RANDOM and ROUND_ROBIN, or a fully qualified class name

Returns:

non-null load balancing policy

getQueryForAllLocalNodes

protected String getQueryForAllLocalNodes()

Gets a SQL query for finding all local nodes regardless which cluster it belongs to. Defaults to QUERY_GET_ALL_NODES.

Returns:

non-null SQL query

getQueryForClusterLocalNodes

protected String getQueryForClusterLocalNodes()

Gets a parameterized SQL query for finding all local nodes of a specific cluster. Defaults to QUERY_GET_CLUSTER_NODES.

Returns:

non-null SQL query

getQueryForNonLocalNodes

protected String getQueryForNonLocalNodes()

Gets a parameterized SQL query for finding all non-local nodes in one or many clusters. Defaults to QUERY_GET_OTHER_NODES.

Returns:

non-null SQL query

get

protected final ClickHouseNode get(ClickHouseNodes manager)

Gets next node available in the list.

Parameters:

manager - managed nodes

Returns:

next node

get

protected ClickHouseNode get(ClickHouseNodes manager,
                             ClickHouseNodeSelector t)

Gets next node available in the list according to the given node selector.

Parameters:

manager - managed nodes

t - node selector

Returns:

next node

Throws:

IllegalArgumentException - when no node available to use

suggest

protected ClickHouseNode suggest(ClickHouseNodes manager,
                                 ClickHouseNode server,
                                 Throwable failure)

Sugguests a new node as replacement of the given one in order to recover from the failure, which is usually a connection error.

Parameters:

manager - non-null manager for finding suitable node

server - non-null server to replace

failure - non-null recoverable exception

Returns:

new server, or the exact same server from input when running out of options

update

protected void update(ClickHouseNodes manager,
                      ClickHouseNode node,
                      ClickHouseNode.Status status)

Updates node status to one of ClickHouseNode.Status.

Parameters:

manager - non-null node manager

node - non-null node to update

status - non-null status of the node

getScheduler

protected ScheduledExecutorService getScheduler()

Gets scheduled executor service for auto discovery and health check.

Returns:

scheduled executor service, null will turn off auto discovery and health check

schedule

protected ScheduledFuture<?> schedule(ScheduledFuture<?> current,
                                      Runnable task,
                                      long interval)

Schedules the task to run only when current run has been completed/cancelled or does not exist.

Parameters:

current - current task, which could be null or in running status

task - scheduled task to run

interval - interval between each run

Returns:

future object representing the scheduled task, could be null when no scheduler available(getScheduler() returns null)

toString

public String toString()

Overrides:

toString in class Object