Connector Guides › Java Connector Server Implementation Guide › Configuring Java CS › Connector Configuration File › Connector Pool Configuration › Retry Configuration

Retry Configuration

You can configure the Exception Map setting to contain groups of exception messages that require special handling (and optionally associated retry delay and retry count settings).

In particular, the JDBC connector defines entries for exceptions signifying these conditions which drive retrying when connections to the endpoint experience problems:

• Stale—The connection to the endpoint has become stale and is reestablished immediately.
• Retriable—The connection to the endpoint has encountered what can be a transient soft failure, in which case a retry loop is started with the configured count and delay. If the count is exhausted before connectivity is restored, then the current request is considered to have suffered a hard failure which is reported to the Java CS client.
• Busy—The endpoint has reported it is too busy to complete a request in which case a retry loop is started with a separate retry delay and count settings. For example, the MSSQL and Ingres databases both report deadlock exceptions when they are unable to complete processing of a transaction within a certain time interval. The delay and recount settings are typically much longer than the Retriable case.

In addition to these triggering exceptions, each ExceptionRetryGroup has associated resilientDelay and resilientMaxRetries settings which specify how many retry attempts are required when a matching exception is encountered, and the delay between each attempt.