Package org.apache.solr.client.solrj.impl

Class CloudSolrClient

java.lang.Object

org.apache.solr.client.solrj.SolrClient

org.apache.solr.client.solrj.impl.CloudSolrClient

All Implemented Interfaces:

Closeable, Serializable, AutoCloseable

Direct Known Subclasses:

CloudHttp2SolrClient

public abstract class CloudSolrClient
extends SolrClient

A SolrClient that routes requests to ideal nodes, including splitting update batches to the correct shards. It uses LBSolrClient as well, thus offering fail-over abilities if a core or node becomes unavailable. It's able to know where to route requests due to its knowledge of the SolrCloud "cluster state".

See Also:

• Serialized Form

Nested Class Summary

Nested Classes

Modifier and Type	Class	Description
static class	CloudSolrClient.Builder	Constructs CloudSolrClient instances from provided configuration.
static class	CloudSolrClient.RouteException
static class	CloudSolrClient.RouteResponse<T extends LBSolrClient.Req>
protected static class	CloudSolrClient.StateCache

Nested classes/interfaces inherited from class org.apache.solr.client.solrj.SolrClient

SolrClient.SolrClientFunction<C extends SolrClient,R>

Field Summary

Fields

Modifier and Type	Field	Description
protected final CloudSolrClient.StateCache	collectionStateCache
protected long	retryExpiryTimeNano
static final String	STATE_VERSION

Fields inherited from class org.apache.solr.client.solrj.SolrClient

defaultCollection

Constructor Summary

Constructors

Modifier	Constructor	Description
protected	CloudSolrClient(boolean updatesToLeaders, boolean parallelUpdates, boolean directUpdatesToLeadersOnly)
protected	CloudSolrClient(boolean updatesToLeaders, boolean parallelUpdates, boolean directUpdatesToLeadersOnly, int stateRefreshThreads)

Method Summary

Modifier and Type	Method	Description
void	close()
CloudSolrClient.RouteResponse	condenseResponse(NamedList<?> response, int timeMillis)
protected <T extends CloudSolrClient.RouteResponse<?>> T	condenseResponse(NamedList<?> response, int timeMillis, Supplier<T> supplier)
void	connect()	Deprecated. Call ClusterStateProvider.getLiveNodes() instead.
void	connect(long duration, TimeUnit timeUnit)	Deprecated.
protected Map<String,? extends LBSolrClient.Req>	createRoutes(UpdateRequest updateRequest, ModifiableSolrParams routableParams, DocCollection col, DocRouter router, Map<String,List<String>> urlMap, String routeField)
ClusterState	getClusterState()	Deprecated. problematic as a 'get' method, since one implementation will do a remote request each time this is called, potentially return lots of data that isn't even needed.
abstract ClusterStateProvider	getClusterStateProvider()
protected DocCollection	getDocCollection(String collection, Integer expectedVersion)
protected abstract LBSolrClient	getLbClient()
int	getMinAchievedReplicationFactor(String collection, NamedList<?> resp)	Useful for determining the minimum achieved replication factor across all shards involved in processing an update request, typically useful for gauging the replication factor of a batch.
ResponseParser	getParser()
RequestWriter	getRequestWriter()
protected CloudSolrClient.RouteException	getRouteException(SolrException.ErrorCode serverError, NamedList<Throwable> exceptions, Map<String,? extends LBSolrClient.Req> routes)
Map<String,Integer>	getShardReplicationFactor(String collection, NamedList<?> resp)	Walks the NamedList response after performing an update request looking for the replication factor that was achieved in each shard involved in the request.
protected int	getStateRefreshParallelism()	Visible for tests so they can assert the configured refresh parallelism.
boolean	isDirectUpdatesToLeadersOnly()	If true, this client has been configured such that "direct updates" will only be sent to the current leader of the corresponding shard, and will not be retried with other replicas.
boolean	isParallelUpdates()	Gets whether direct updates are sent in parallel
boolean	isUpdatesToLeaders()	If true, this client has been configured such that it will generally prefer to send SolrRequest.SolrRequestType.UPDATE requests to a shard leader, if and only if UpdateRequest.isSendToLeaders() is also true.
NamedList<Object>	request(SolrRequest<?> request, String collection)	Execute a request against a Solr server for a given collection
protected NamedList<Object>	requestWithRetryOnStaleState(SolrRequest<?> request, int retryCount, List<String> inputCollections, boolean skipStateVersion, Map<String,CompletableFuture<DocCollection>> pendingRefreshes, boolean waitedForRefresh)	As this class doesn't watch external collections on the client side, there's a chance that the request will fail due to cached stale state, which means the state must be refreshed from ZK and retried.
protected NamedList<Object>	sendRequest(SolrRequest<?> request, List<String> inputCollections)
protected boolean	wasCommError(Throwable t)	Is this a communication error? We will retry if so.

Methods inherited from class org.apache.solr.client.solrj.SolrClient

add, add, add, add, add, add, add, add, add, add, addBean, addBean, addBean, addBean, addBeans, addBeans, addBeans, addBeans, addBeans, addBeans, commit, commit, commit, commit, commit, commit, deleteById, deleteById, deleteById, deleteById, deleteById, deleteById, deleteById, deleteById, deleteByQuery, deleteByQuery, deleteByQuery, deleteByQuery, getById, getById, getById, getById, getById, getById, getById, getById, getContext, getDefaultCollection, optimize, optimize, optimize, optimize, optimize, optimize, ping, ping, query, query, query, query, queryAndStreamResponse, queryAndStreamResponse, queryAndStreamResponse, request, rollback, rollback

Methods inherited from class java.lang.Object

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

Field Details

STATE_VERSION

public static final String STATE_VERSION

See Also:

• Constant Field Values

retryExpiryTimeNano

protected long retryExpiryTimeNano

collectionStateCache

protected final CloudSolrClient.StateCache collectionStateCache

Constructor Details

CloudSolrClient

protected CloudSolrClient(boolean updatesToLeaders,
 boolean parallelUpdates,
 boolean directUpdatesToLeadersOnly)

CloudSolrClient

protected CloudSolrClient(boolean updatesToLeaders,
 boolean parallelUpdates,
 boolean directUpdatesToLeadersOnly,
 int stateRefreshThreads)

Method Details

getLbClient

protected abstract LBSolrClient getLbClient()

getClusterStateProvider

public abstract ClusterStateProvider getClusterStateProvider()

getClusterState

@Deprecated
public ClusterState getClusterState()

Deprecated.

problematic as a 'get' method, since one implementation will do a remote request each time this is called, potentially return lots of data that isn't even needed.

wasCommError

protected boolean wasCommError(Throwable t)

Is this a communication error? We will retry if so.

close

public void close()

getParser

public ResponseParser getParser()

getRequestWriter

public RequestWriter getRequestWriter()

isParallelUpdates

public boolean isParallelUpdates()

Gets whether direct updates are sent in parallel

connect

@Deprecated
public void connect()

Deprecated.

Call ClusterStateProvider.getLiveNodes() instead.

Connect to the zookeeper ensemble. This is an optional method that may be used to force a connection before any other requests are sent.

connect

@Deprecated
public void connect(long duration,
 TimeUnit timeUnit)
             throws TimeoutException,
InterruptedException

Deprecated.

Connect to a cluster. If the cluster is not ready, retry connection up to a given timeout.

Parameters:

duration - the timeout

timeUnit - the units of the timeout

Throws:

TimeoutException - if the cluster is not ready after the timeout

InterruptedException - if the wait is interrupted

getRouteException

protected CloudSolrClient.RouteException getRouteException(SolrException.ErrorCode serverError,
 NamedList<Throwable> exceptions,
 Map<String,? extends LBSolrClient.Req> routes)

createRoutes

protected Map<String,? extends LBSolrClient.Req> createRoutes(UpdateRequest updateRequest,
 ModifiableSolrParams routableParams,
 DocCollection col,
 DocRouter router,
 Map<String,List<String>> urlMap,
 String routeField)

condenseResponse

protected <T extends CloudSolrClient.RouteResponse<?>>
T condenseResponse(NamedList<?> response,
 int timeMillis,
 Supplier<T> supplier)

condenseResponse

public CloudSolrClient.RouteResponse condenseResponse(NamedList<?> response,
 int timeMillis)

request

public NamedList<Object> request(SolrRequest<?> request,
 String collection)
                          throws SolrServerException,
IOException

Description copied from class: SolrClient

Execute a request against a Solr server for a given collection

Specified by:

request in class SolrClient

Parameters:

request - the request to execute

collection - the collection to execute the request against

Returns:

a NamedList containing the response from the server

Throws:

SolrServerException - if there is an error on the server

IOException - If there is a low-level I/O error.

requestWithRetryOnStaleState

protected NamedList<Object> requestWithRetryOnStaleState(SolrRequest<?> request,
 int retryCount,
 List<String> inputCollections,
 boolean skipStateVersion,
 Map<String,CompletableFuture<DocCollection>> pendingRefreshes,
 boolean waitedForRefresh)
                                                  throws SolrServerException,
IOException

As this class doesn't watch external collections on the client side, there's a chance that the request will fail due to cached stale state, which means the state must be refreshed from ZK and retried.

Throws:

SolrServerException

IOException

sendRequest

protected NamedList<Object> sendRequest(SolrRequest<?> request,
 List<String> inputCollections)
                                 throws SolrServerException,
IOException

Throws:

SolrServerException

IOException

isUpdatesToLeaders

public boolean isUpdatesToLeaders()

If true, this client has been configured such that it will generally prefer to send SolrRequest.SolrRequestType.UPDATE requests to a shard leader, if and only if UpdateRequest.isSendToLeaders() is also true. If false, then this client has been configured to obey normal routing preferences when dealing with SolrRequest.SolrRequestType.UPDATE requests.

See Also:

• isDirectUpdatesToLeadersOnly()

isDirectUpdatesToLeadersOnly

public boolean isDirectUpdatesToLeadersOnly()

If true, this client has been configured such that "direct updates" will only be sent to the current leader of the corresponding shard, and will not be retried with other replicas. This method has no effect if isUpdatesToLeaders() or UpdateRequest.isSendToLeaders() returns false.

A "direct update" is any update that can be sent directly to a single shard, and does not need to be broadcast to every shard. (Example: document updates or "delete by id" when using the default router; non-direct updates are things like commits and "delete by query").

NOTE: If a single UpdateRequest contains multiple "direct updates" for different shards, this client may break the request up and merge the responses.

Returns:

true if direct updates are sent to shard leaders only

getStateRefreshParallelism

protected int getStateRefreshParallelism()

Visible for tests so they can assert the configured refresh parallelism.

getDocCollection

protected DocCollection getDocCollection(String collection,
 Integer expectedVersion)
                                  throws SolrException

Throws:

SolrException

getMinAchievedReplicationFactor

public int getMinAchievedReplicationFactor(String collection,
 NamedList<?> resp)

Useful for determining the minimum achieved replication factor across all shards involved in processing an update request, typically useful for gauging the replication factor of a batch.

getShardReplicationFactor

public Map<String,Integer> getShardReplicationFactor(String collection,
 NamedList<?> resp)

Walks the NamedList response after performing an update request looking for the replication factor that was achieved in each shard involved in the request. For single doc updates, there will be only one shard in the return value.