Sending Guaranteed Messages
When sending a Guaranteed message, client applications must consider the following factors:
- When using the Java RTO, C, or .NET APIs, whether you want to send messages using a blocking or non-blocking mode. See Blocking and Non-Blocking Network I/O in the Java RTO, C, and .NET APIs.
- The number of messages to publish per send invocation. See Sending One Message at a Time or Sending Multiple Messages at Once.
- Whether you want to send messages that require a reply from the receiver. See Publishing Messages that Request Replies.
-
If you are publishing in a network environment that experiences higher than usual delays, you might need to adjust the Guaranteed message publish window size (see Adjusting the Guaranteed Message Publish Window Size ) or acknowledgment timer (see Adjusting the Guaranteed Message Publish ACK Timer).
Related Samples
For examples of how to publish and receive Guaranteed messages, see the simpleFlowToQueue
sample for the Java RTO, C, and .NET APIs, and the QueueProducer
sample for the JavaScript and Node.js APIs.
Blocking and Non-Blocking Network I/O in the Java RTO, C, and .NET APIs
The Java RTO, C, and .NET APIs can use a blocking or non‑blocking network I/O model for send operations. This is controlled by the send blocking session property. By default, this property is set to true or ENABLE, so that a blocking mode is used.
When the send blocking session property is set to false
or DISABLE
, a non-blocking mode is used. If the OS/network cannot send the whole message without blocking the calling thread until more buffer space is available, the client application gets a WOULD_BLOCK
return code. When a WOULD_BLOCK
return code is received, the message is not sent and the client application must wait for a CAN_SEND
session event, which indicates that the network is then ready for the send.
Adjusting the Guaranteed Message Publish Window Size
The Guaranteed message publish window size can be adjusted, if necessary, to better suit your messaging environment. The valid range is 1 through 255.
Although the default value has been proven to provide good performance for most LAN situations, increasing the value might help in a situation where messages are being published over a WAN or a network with higher delay (possibly incurring event broker/switch induced latencies), and the window is experiencing flow control situations.
Windowed acknowledgment properties have no effect for Direct messages.
PubSub+ Messaging API | Use |
---|---|
Java RTO |
SessionHandle.PROPERTIES.PUB_WINDOW_SIZE |
C |
SOLCLIENT_SESSION_PROP_PUB_WINDOW_SIZE |
.NET |
SessionProperties.ADPublishWindowSize |
JavaScript and Node.js |
solace.MessagePublisherProperties.windowSize where the |
Adjusting the Guaranteed Message Publish ACK Timer
The Guaranteed message publish acknowledgment timer sets the amount of time (in milliseconds) that the publishing API will wait for an acknowledgment from the event broker before resending. Guaranteed messages not acknowledged within the publish acknowledgment time are retransmitted automatically by the API.
Although the default Guaranteed message publish acknowledgment timer value has been proven to provide good performance, it can be adjusted, if necessary.
Windowed acknowledgment properties have no effect for Direct messages.
PubSub+ Messaging API | Use |
---|---|
Java RTO |
SessionHandle.PROPERTIES.PUB_ACK_TIMER |
C |
SOLCLIENT_SESSION_PROP_PUB_ACK_TIMER |
.NET |
SessionProperties.ADPublishAckTimerInMsecs |
JavaScript and Node.js |
solace.MessagePublisherProperties.acknowledgeTimeoutInMsecs where the |
Sending One Message at a Time
To publish a single message per API call, call one of the send methods or functions listed below.
PubSub+ Messaging API | Use |
---|---|
Java RTO |
SessionHandle.send(...) |
C |
solClient_session_sendMsg(...) |
.NET |
ISession.Send(...) |
JavaScript and Node.js |
solace.Session.send(...) |
Sending Multiple Messages at Once
A group of Direct or Guaranteed messages can be sent through a single API call. This allows messages to be sent in a batch or a vector. The messages to be sent are specified as an array; up to fifty messages can be sent through the call.
When batch-sending messages through a send-multiple API call, the same Delivery mode (for example, Direct or persistent) should be set for all of the messages (see Delivery Mode).
Java RTO, C, JavaScript, Node.js, and .NET APIs
When using the Java RTO, C, JavaScript, Node.js, or .NET APIs, an array of messages must be passed in. Message destinations are not set through a send method, rather they are set as a property for each message.
When a blocking mode is configured for the session through the session properties, the call blocks until all messages can be accepted by the messaging API. If a non‑blocking mode is configured, and the API cannot accept all messages to publish, SOLCLIENT_WOULD_BLOCK
is returned, and the number of messages that were accepted is returned in the in_out_howManySent
, numberOfMessagesWritten
, or messagesSent
parameter. The application can reattempt to send the messages that were not accepted or take another action.
When Guaranteed messages are sent, messages are transmitted serially.
PubSub+ Messaging API | Use |
---|---|
Java RTO |
SessionHandle.send(...) |
C |
solClient_session_sendMultipleMsg(...) |
.NET |
ISession.Send(...) |
JavaScript and Node.js |
Not applicable |
Related Samples
For examples of how to publish a group of messages, see the PerfTest
sample for the appropriate messaging API.
Message Correlation
When using Guaranteed messaging, a correlation key or tag is used to correlate a message with its ACK or NACK. For the Java RTO and .NET APIs, the correlation key is an object that is passed back to the client during the event broker ACK or NACK. For the C API, the correlation tag is a pointer that is passed back to the client during the event broker ACK or NACK.
PubSub+ Messaging API | Use |
---|---|
Java RTO |
To set a correlation key on a message:
The correlation key is passed back to the client through the following session events:
|
C |
To set a correlation tag on a message:
The correlation tag is passed back to the client in the session event callback for the following session events:
|
.NET |
To set the correlation key property of a message:
The correlation key is passed back to the client in the
|
JavaScript and Node.js |
To set the correlation key property of a message:
The correlation key is passed back to the client in the listeners for the following events:
|
Related Samples
For an example of how to use correlation keys, see the adPubAck
sample for the appropriate messaging API.
For the JavaScript and Node.js APIs, see the ConfirmedPublish
sample.