Client Settings for the Java SDK

The ClusterEnvironment class enables you to configure Java SDK options for bootstrapping, timeouts, reliability, and performance.

The Environment Builder

Most client settings are related to the ClusterEnvironment. Because ClusterEnvironment is an immutable class, you need to configure it by using its embedded Builder class. It is designed to apply the builder arguments in a fluent fashion and then create the ClusterEnvironment at the very end.

Creating a cluster with custom settings
ClusterEnvironment env = ClusterEnvironment.builder()
    // [Customize client settings here]

// Create a cluster using the custom client settings.
Cluster cluster = Cluster.connect(connectionString, ClusterOptions
    .clusterOptions(username, password)

// [Your code to interact with the cluster]

// Shut down gracefully.

Config Builders

Environment settings are grouped into categories, with one builder class per category. These builders all work in the same way, which we’ll illustrate by using timeout settings as an example.

Timeout settings are configured using an instance of TimeoutConfig.Builder. The usual way to create an instance is to call a static factory method on the TimeoutConfig class. The method TimeoutConfig.builder() returns a builder with default settings. There are also static factory methods that create a builder and set a configuration value in one step. For example, instead of TimeoutConfig.builder().kvTimeout(…​) you can write simply TimeoutConfig.kvTimeout(…​).

Creating a new timeout config builder
ClusterEnvironment env = ClusterEnvironment.builder()

Another way to obtain an instance of a config builder is to borrow it from the cluster environment builder. This technique may be useful if you’re configuring the environment in stages and wish to preserve values set by a previous stage.

Borrowing the existing timeout config builder
ClusterEnvironment.Builder envBuilder = ClusterEnvironment.builder();
envBuilder.timeoutConfig() // returns a TimeoutConfig.Builder
ClusterEnvironment env =;

The name of the cluster environment builder method for getting and setting each config builder always matches the name of the config class. For example, TimeoutConfig is set via the environment builder’s timeoutConfig method, IoConfig is set via the ioConfig method, and so on.

System Properties

Many client settings may also be configured by setting a Java system property. If a system property is set, it always takes precedence over the builder setting.

Configuration via system property
System.setProperty("com.couchbase.env.timeout.kvTimeout", "10s"); (1)
System.setProperty("com.couchbase.env.timeout.queryTimeout", "15s");

ClusterEnvironment environment = ClusterEnvironment.builder()
    .timeoutConfig(TimeoutConfig.kvTimeout(Duration.ofSeconds(5))) (2)
1 When specifying durations, s stands for seconds. Other valid qualifiers are ns for nanoseconds, us for microseconds, ms for milliseconds, and m for minutes.
2 The kvTimeout value specified via TimeoutConfig.Builder is overridden by the system property. In this example the actual kvTimeout is 10 seconds, and the queryTimeout is 15 seconds.
System property names starting with com.couchbase.env are paths in a Java object graph rooted at the environment builder. Setting the property com.couchbase.env.timeout.kvTimeout tells the SDK to invoke envBuilder.timeoutConfig().kvTimeout(…​) using reflection.

Configuration Options

The following tables cover all possible configuration options and explain their usage and default values. The tables categorize the options into groups for bootstrapping, timeout, reliability, performance, and advanced options.

Security Options

By default the client will connect to Couchbase Server using an unencrypted connection. If you are using the Enterprise Edition, it’s possible to secure the connection using TLS.

Template for configuring security settings
ClusterEnvironment env = ClusterEnvironment.builder()
Unless you set enableTls to true, none of the other security settings in this section have any effect.
Name: Enabling Secure Connections

Builder Method: SecurityConfig.enableTls(boolean)

Default: false

System Property:

Set this to true to encrypt all communication between the client and server using TLS. This feature requires the Enterprise Edition of Couchbase Server 3.0 or later. If TLS is enabled you must also specify the trusted certificates by calling exactly one of trustCertificate, trustCertificates, or trustManagerFactory. Please see the Managing Connections section for more details on how to set it up properly.

Name: Disabling Native TLS Provider

Builder Method: SecurityConfig.enableNativeTls(boolean)

Default: true

System Property:

When TLS is enabled, the client will by default use an optimized native TLS provider if one is available. If for some reason you need to disable the native provider and use the JDK’s portable provider instead, set this to false. If enableTls is false then enableNativeTls has no effect.

Name: TLS Certificate Location

Builder Method: SecurityConfig.trustCertificate(Path)

Default: N/A

System Property:

Path to a file containing a single X.509 certificate to trust as a Certificate Authority when establishing secure connections. See the Connection Management section for more details on how to set it up properly.

Name: TLS Certificates

Builder Method: SecurityConfig.trustCertificates(List<X509Certificate>)

Default: N/A

System Property: N/A

If you wish to trust more than one certificate, or prefer to load the certificate yourself, then call this method to specify the certificates to trust as Certificate Authorities when establishing secure connections. See the Connection Management section for more details on how to set it up properly.

Name: Custom TLS Trust Manager Factory

Builder Method: SecurityConfig.trustManagerFactory(TrustManagerFactory)

Default: N/A

System Property: N/A

As an alternative to specifying the certificates to trust, you can specify a custom TrustManagerFactory to use when establishing secure connections. See the Connection Management section for more details on how to set it up properly.

I/O Options

I/O settings are represented by the Java class IoConfig. The associated ClusterEnvironement.Builder method is called ioConfig.

Template for configuring I/O settings
ClusterEnvironment env = ClusterEnvironment.builder()
Name: DNS SRV Enabled

Builder Method: IoConfig.enableDnsSrv(boolean)

Default: true

System Property:

Gets the bootstrap node list from a DNS SRV record. See the Connection Management section for more information on how to use it properly.

Name: Mutation Tokens Enabled

Builder Method: IoConfig.mutationTokensEnabled(boolean)

Default: true

System Property:

Mutation tokens allow enhanced durability requirements as well as advanced N1QL querying capabilities. Set this to false if you do not require these features and wish to avoid the associated overhead.

Name: Network Resolution

Builder Method: IoConfig.networkResolution(NetworkResolution)

Default: auto

System Property:

The system property value should be one of auto, default, or external (lower case).

Each node in the Couchbase Server cluster might have multiple addresses associated with it. For example, a node might have one address that should be used when connecting from inside the same virtual network environment where the server is running, and a second address for connecting from outside the server’s network environment.

By default the client will use a simple matching heuristic to determine which set of addresses to use (it will select the set of addresses that contains a seed node’s host and port).

If you wish to override the heuristic, you can set this value to default if the client is running in the same network as the server, or external if the client is running in a different network.

Name: Capture Traffic

Builder Method: IoConfig.captureTraffic(ServiceType…​)

Default: capture is disabled

System Property:

Multiple services may be specified in the system property value using a comma-delimited list such as KV,QUERY. To enable capture for all services, set the value of the system property to an empty string.

Call this method to log all traffic to the specified services. If no services are specified, traffic to all services is captured.

Name: Socket Keepalive

Builder Method: IoConfig.enableTcpKeepAlives(boolean)

Default: true

System Property:

If enabled, the client periodically sends a TCP keepalive to the server to prevent firewalls and other network equipment from dropping idle TCP connections.

Name: Socket Keepalive Interval

Builder Method: IoConfig.tcpKeepAliveTime(Duration)

Default: 60s

System Property:

The idle time after which a TCP keepalive gets fired. (This setting has no effect if enableTcpKeepAlives is false.)

This setting only propagates to the OS on Linux when the epoll transport is used. On all other platforms, the OS-configured time is used (and you need to tune it there if you want to override the default interval).
Name: Key/Value Endpoints per Node

Builder Method: IoConfig.numKvConnections(int)

Default: 1

System Property:

The number of actual endpoints (sockets) to open per node in the cluster against the Key/Value service. By default, for every node in the cluster one socket is opened where all traffic is pushed through. That way the SDK implicitly benefits from network batching characteristics when the workload increases. If you suspect based on profiling and benchmarking that the socket is saturated you can think about slightly increasing it to have more "parallel pipelines". This might be especially helpful if you need to push large documents through it. The recommendation is keeping it at 1 unless there is other evidence.

Durable Write operations with Couchbase Server 6.5 and above require up to 16 KV Endpoints per node, for most efficient operation, unless the environment dictates something a little lower.
Name: Max HTTP Endpoints per Service per Node

Builder Method: IoConfig.maxHttpConnections(int)

Default: 12

System Property:

Each service (except the Key/Value service) has a separate dynamically sized pool of HTTP connections for issuing requests. This setting puts an upper bound on the number of HTTP connections in each pool.

Name: Idle HTTP Connection Timeout

Builder Method: IoConfig.idleHttpConnectionTimeout(Duration)

Default: 30s

System Property:

The length of time an HTTP connection may remain idle before it is closed and removed from the pool. Durations longer than 50 seconds are not recommended, since some services have a 1 minute server side idle timeout.

Name: Config Poll Interval

Builder Method: IoConfig.configPollInterval(Duration)

Default: 2.5s

System Property:

The interval at which the client fetches cluster topology information in order to proactively detect changes.

Circuit Breaker Options

Circuit breakers are a tool for preventing cascading failures.

When a circuit is closed, requests are sent to the server as normal. If too many requests fail within a certain time window, the breaker opens the circuit, preventing requests from going through.

When a circuit is open, any requests to the service immediately fail without the client even talking to the server. After a "sleep delay" elapses, the next request is allowed to go through the to the server. This trial request is called a "canary."

Each service has an associated circuit breaker which may be configured independently of the others. The IoConfig builder has methods for configuring the circuit breakers of each service.

Template for configuring circuit breaker settings
ClusterEnvironment env = ClusterEnvironment.builder()

The corresponding system properties would be:

For the other services, replace kv with query, view, search, analytics, or manager.

The properties of a circuit breaker are described below.


Default: true

Enables or disables this circuit breaker.

If this property is set to false, then the circuit breaker is not used and all other properties are ignored.


Default: 20

The volume threshold defines how many operations must be in the window before the threshold percentage can be meaningfully calculated.


Default: 50

The percentage of operations in a window that may fail before the circuit is opened. The value is an integer in the range [0,100].


Default: 5s

The delay between when the circuit opens and when the canary is tried.


Default: 1m

How long the window is in which the number of failed ops are tracked in a rolling fashion.

Timeout Options

The default timeout values are suitable for most environments, and should be adjusted only after profiling the expected latencies in your deployment environment. If you get a timeout exception, it may be a symptom of another issue; increasing the timeout duration is sometimes not the best long-term solution.

Most timeouts can be overridden on a per-operation basis (for example, by passing a custom options block to a "get" or "query" method). The values set here are used as the defaults when no per-operation timeout is specified. See setting duration values under System Properties.

Timeout settings are represented by the Java class TimeoutConfig. The associated ClusterEnvironement.Builder method is called timeoutConfig.

Template for configuring timeouts
ClusterEnvironment env = ClusterEnvironment.builder()

Timeout Options Reference

Name: Key-Value Timeout

Builder Method: TimeoutConfig.kvTimeout(Duration)

Default: 2.5s — but see TIP, below

System Property: com.couchbase.env.timeout.kvTimeout

The Key/Value default timeout is used on operations which are performed on a specific key if not overridden by a custom timeout. This includes all commands like get(), getFromReplica() and all mutation commands, but does not include operations that are performed with enhanced durability requirements.

Durable Write operations have their own timeout setting, kvDurableTimeout, see below.
Name: Key-Value Durable Operation Timeout

Builder Method: TimeoutConfig.kvDurableTimeout(Duration)

Default: 10s

System Property: com.couchbase.env.timeout.kvDurableTimeout

Key/Value operations with enhanced durability requirements may take longer to complete, so they have a separate default timeout.

The kvDurableTimeout property is not part of the stable API and may change or be removed at any time.
Name: View Timeout

Builder Method: TimeoutConfig.viewTimeout(Duration)

Default: 75s

System Property: com.couchbase.env.timeout.viewTimeout

The View timeout is used on view operations if not overridden by a custom timeout. Note that it is set to such a high timeout compared to key/value since it can affect hundreds or thousands of rows. Also, if there is a node failure during the request the internal cluster timeout is set to 60 seconds.

Name: Query Timeout

Builder Method: TimeoutConfig.queryTimeout(Duration)

Default: 75s

System Property: com.couchbase.env.timeout.queryTimeout

The Query timeout is used on all N1QL query operations if not overridden by a custom timeout. Note that it is set to such a high timeout compared to key/value since it can affect hundreds or thousands of rows.

Name: Search Timeout

Builder Method: TimeoutConfig.searchTimeout(Duration)

Default: 75s

System Property: com.couchbase.env.timeout.searchTimeout

The Search timeout is used on all FTS operations if not overridden by a custom timeout. Note that it is set to such a high timeout compared to key/value since it can affect hundreds or thousands of rows.

Name: Analytics Timeout

Builder Method: TimeoutConfig.analyticsTimeout(Duration)

Default: 75s

System Property: com.couchbase.env.timeout.analyticsTimeout

The Analytics timeout is used on all Analytics query operations if not overridden by a custom timeout. Note that it is set to such a high timeout compared to key/value since it can affect hundreds or thousands of rows.

Name: Connect Timeout

Builder Method: TimeoutConfig.connectTimeout(Duration)

Default: 10s

System Property: com.couchbase.env.timeout.connectTimeout

The connect timeout is used when a Bucket is opened and if not overridden by a custom timeout. If you feel the urge to change this value to something higher, there is a good chance that your network is not properly set up. Connecting to the server should in practice not take longer than a second on a reasonably fast network.

Name: Disconnect Timeout

Builder Method: TimeoutConfig.disconnectTimeout(Duration)

Default: 10s

System Property: com.couchbase.env.timeout.disconnectTimeout

The disconnect timeout is used when a Cluster is disconnected and if not overridden by a custom timeout. A timeout is applied here always to make sure that your code does not get stuck at shutdown. The default should provide enough time to drain all outstanding operations properly, but make sure to adapt this timeout to fit your application requirements.

Name: Management Timeout

Builder Method: TimeoutConfig.managementTimeout(Duration)

Default: 75s

System Property: com.couchbase.env.timeout.managementTimeout

The management timeout is used on all cluster management APIs (BucketManager, UserManager, CollectionManager, QueryIndexManager, etc.) if not overridden by a custom timeout. The default is quite high because some operations (such as flushing a bucket, for example) might take a long time.

Diagnostics Options

The client can periodically log diagnostic information about server connectivity.

Diagnostics settings are represented by the Java class DiagnosticsConfig. The associated ClusterEnvironement.Builder method is called diagnosticsConfig.

Template for configuring diagnostics settings
ClusterEnvironment env = ClusterEnvironment.builder()
Name: Enabling Diagnostics

Builder Method: DiagnosticsConfig.enabled(boolean)

Default: false

System Property: com.couchbase.env.diagnostics.enabled

If enabled, the client periodically logs diagnostic information.

Name: Diagnostics Interval

Builder Method: DiagnosticsConfig.emitInterval(Duration)

Default: 30m

System Property: com.couchbase.env.diagnostics.emitInterval

The interval at which diagnostic information is logged.

Compression Options

The client can optionally compress documents before sending them to Couchbase Server.

Compression settings are represented by the Java class CompressionConfig. The associated ClusterEnvironement.Builder method is called compressionConfig.

Template for configuring compression settings
ClusterEnvironment env = ClusterEnvironment.builder()
Name: Enabling Compression

Builder Method: CompressionConfig.enabled

Default: true

System Property: com.couchbase.env.compression.enabled(boolean)

If enabled, the client will compress documents before they are sent to Couchbase Server. If this is set to false, the other compression settings have no effect.

Name: Document Minimum Size

Builder Method: CompressionConfig.minSize(int)

Default: 32

System Property: com.couchbase.env.compression.minSize

Size in bytes. Documents smaller than this size are never compressed.

Name: Document Minimum Compressibility

Builder Method: CompressionConfig.minRatio(double)

Default: 0.83

System Property: com.couchbase.env.compression.minRatio

A floating point value between 0 and 1. Specifies how "compressible" a document must be in order for the compressed form to be sent to the server.

Increasing the value allows compression to be used with less-compressible documents.

If the compressed document size divided by the uncompressed document size is greater than this value, then the uncompressed version of the document will be sent to Couchbase Server instead of the compressed version.

For example, with a minRatio of 0.83, compression will only be used if the size of the compressed document is less than 83% of the uncompressed document size.

General Options

The settings in this category apply to the client in general. They are configured directly on the ClusterEnvironment.Builder.

Template for configuring general settings
ClusterEnvironment env = ClusterEnvironment.builder()
Name: Retry Strategy

Builder Method: retryStrategy(RetryStrategy)

Default: BestEffortRetryStrategy.INSTANCE

System Property: N/A

The client’s default retry strategy.

A retry strategy decides whether a failed operation should be retried. Implementing a custom strategy is fairly advanced, so the SDK ships with two out of the box: BestEffortRetryStrategy and FailFastRetryStrategy.

The "best effort" strategy will retry the operation until it either succeeds or the timeout expires. The "fail fast" strategy will immediately report the failure to your application, giving you more control over how and when to retry.

Most client operations that accept an options block allow for overriding the default strategy as one of the options.

See the advanced section in the documentation on more specific information on retry strategies and failure management.

Name: JSON Serializer

Builder Method: jsonSerializer(JsonSerializer)

Default: see below

System Property: N/A

The JSON serializer handles the conversion between JSON and Java objects.

If Jackson is present in the class path, the default serializer will be an instance of JacksonJsonSerializer using a default ObjectMapper.

To create a serializer backed by a custom ObjectMapper, call JacksonJsonSerializer.create and pass in your custom mapper.

If Jackson is not present, the client will fall back to using an unspecified default serializer. (Actually, it will use a repackaged version of Jackson, but this is an implementation detail you should not depend on.)

Name: Transcoder

Builder Method: transcoder(Transcoder)

Default: JsonTranscoder

System Property: N/A

The transcoder is responsible for converting KV binary packages to and from Java objects.

The default transcoder assumes you are working with JSON documents. It uses the configured jsonSerializer to convert between JSON and Java objects. When writing documents it sets the appropriate flags to indicate the document content is JSON.

The transcoder configured here is just the default; it can be overridden on a per-operation basis.

Name: Request Tracer

Builder Method: requestTracer(RequestTracer)

Default: ThresholdRequestTracer

System Property: N/A

The default tracer logs the slowest requests per service.

Various RequestTracer implementations exist, both as part of the core library and as external modules that can be attached (i.e. for OpenTracing and OpenTelemetry). It is recommended to use those modules and not write your own tracer unless absolutely needed.

When using a non-default tracer, you are responsible for starting and stopping it.
Name: Computation Scheduler

Builder Method: scheduler(Scheduler)

Default: see below

System Property: N/A

This is an advanced setting that should not be modified without good reason.

The scheduler used for all CPU-intensive, non-blocking computations in the core, client, and user space. The default is a scheduler created from Reactor’s Schedulers.newParallel method, with one daemon thread per CPU core. Extra care should be used when changing the scheduler, since many internal components depend on it.

Shutting down the cluster environment will not dispose of a custom scheduler. You are responsible for disposing of it after it is no longer needed.
Name: Event Bus

Builder Method: eventBus(EventBus)

Default: DefaultEventBus

System Property: N/A

This is an advanced setting that should not be modified without good reason.

The event bus implementation used to transport system, performance, and debug events from producers to subscribers. If you provide a custom implementation, double check that it fits with the contract of the event bus as documented.

Shutting down the cluster environment will not stop a custom event bus. You are responsible for stopping it after it is no longer needed.