Because transactions can be executed in a replica database and then wait for a period of time before being propagated to the master database, it is important to consider the effect of making changes to stored procedures that are called by these transactions.

Stored procedures are identified by their call interface (name and parameter list). Therefore, as long the call interface remains the same, you can modify the content of stored procedures and any transactions that are in-flight should still execute successfully when they reach the master database.

However, if you want to change a stored procedure to process different parameters, you should save the stored procedure with a new name, and retain the previous version of the procedure in the master database. This ensures that any transactions that are still on their way to the master database can execute successfully when they arrive.