Package com.google.cloud.grpc

Class GcpManagedChannelOptions.GcpResiliencyOptions.Builder

java.lang.Object

com.google.cloud.grpc.GcpManagedChannelOptions.GcpResiliencyOptions.Builder

Enclosing class:

GcpManagedChannelOptions.GcpResiliencyOptions

public static class GcpManagedChannelOptions.GcpResiliencyOptions.Builder
extends Object

Constructor Summary

Constructors

Constructor	Description
Builder()
Builder(GcpManagedChannelOptions.GcpResiliencyOptions options)

Method Summary

Modifier and Type	Method	Description
GcpManagedChannelOptions.GcpResiliencyOptions	build()
GcpManagedChannelOptions.GcpResiliencyOptions.Builder	disableUnresponsiveConnectionDetection()	Disable unresponsive connection detection.
GcpManagedChannelOptions.GcpResiliencyOptions.Builder	setNotReadyFallback(boolean enabled)	If true, temporarily fallback requests to a ready channel from a channel which is not ready to send a request immediately.
GcpManagedChannelOptions.GcpResiliencyOptions.Builder	withUnresponsiveConnectionDetection(int ms, int numDroppedRequests)	Enable unresponsive connection detection.

Methods inherited from class java.lang.Object

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

Constructor Details

Builder

public Builder()

Builder

public Builder(GcpManagedChannelOptions.GcpResiliencyOptions options)

Method Details

build

public GcpManagedChannelOptions.GcpResiliencyOptions build()

setNotReadyFallback

public GcpManagedChannelOptions.GcpResiliencyOptions.Builder setNotReadyFallback(boolean enabled)

If true, temporarily fallback requests to a ready channel from a channel which is not ready to send a request immediately. The fallback will happen if the pool has another channel in the READY state and that channel has less than maximum allowed concurrent active streams.

withUnresponsiveConnectionDetection

public GcpManagedChannelOptions.GcpResiliencyOptions.Builder withUnresponsiveConnectionDetection(int ms,
 int numDroppedRequests)

Enable unresponsive connection detection.

If an RPC channel fails to receive any RPC message from the server for ms milliseconds and there were numDroppedRequests calls (started after the last response from the server) that resulted in DEADLINE_EXCEEDED then a graceful reconnection of the channel will be performed.

During the reconnection a new subchannel (connection) will be created for new RPCs, and the calls on the old subchannel will still have a chance to complete if the server side responds. When all RPCs on the old subchannel finish the old connection will be closed.

The ms should not be less than the timeout used for the majority of calls. And numDroppedRequests must be > 0.

The logic treats any message from the server almost as a "ping" response. But only calls started after the last response received and ended up in DEADLINE_EXCEEDED count towards numDroppedRequests. Because of that, it may not detect an unresponsive connection if you have long-running streaming calls only.

disableUnresponsiveConnectionDetection

public GcpManagedChannelOptions.GcpResiliencyOptions.Builder disableUnresponsiveConnectionDetection()

Disable unresponsive connection detection.