Spring Transaction Synchronization of JDBC and JMS

This article might be of help Distributed transactions in Spring, with and without XA. I don't think it covers your case specifically - sending message + updating database.

Official Spring Boot repository contain JTA examples that combine JMS with JDBC based on Atomikos, Bitronix or Java EE server JBoss WildFly.

Additionally I also created few examples that are located in my Github repository. This contains also non-Spring Boot (pure Spring) example.

If you are using Local transactionsAnd the usecase is save to database and then send to jms

Then there could be three cases :

1. Exception just after receiving(before DB and JMS)

No problem everything will be rolledback

2. After saving to DB , we have exception

If there is an insert operation,there will be mutiple rows in DB due to retries.With each retry , an insert will be done.And for JMS , message will go to DeadLetterQueue

3. After saving to DB and sending to JMS , we have an exception

   If there is an insert operation,there will be mutiple rows in DB due to retries.With each retry , an insert will be done.And for JMS , message will go to DeadLetterQueue

Now you dont want to use XA, so the solutions could be

1)Check If(message.getJmsRedelivered() {…}

If not , process it

If its redelivered , check if you processed it already

Check if the data is in database based on details in message

Note that re-deliveries are rare so , this check is also rare and there is no overhead

2)If your method is idempotent , then you dont need this check

And regarding XA , XA guarantees that message is delivered only onceAnd synchronize the transaction across multiple resources

But with XA , you have overhead

So if you can manage without XA , it is preferable