Question

In mysql we use @ for user-defined variable, which stays alive until the connection is closed. In java, when multiple threads share one connection pool, while calling a stored procedure concurrently to get rankings:

BEGIN
  SET @rank := 0;
  SELECT @rank := @rank + 1 as rank FROM ...
END

If 2 threads are calling the procedure at same time, without synchronizing @rank, is it possible @rank may return unexpected result?

Is there a better way to handle this situation?

Thanks!

Answer 1

Whether you get “unexpected results” or not depends on what you expect, which is not clear to me from your question.

Are you trying to isolate the variable to a specific connection, or increment it as a shared counter?

The scope of a user-defined variable is limited to a single connection, and each connection will be used by a single thread at any given time, so with your example you should expect the value to be isolated within each connection. Since you are re-initializing the value to 0 in your code, you should not see any residual effects of separate processes that used the connection previously.

However, since you said you are using a stored procedure, I recommend that you declare a local variable inside the stored procedure and use that instead of a user-defined variable.

The local variable will be scoped to the stored procedure, so you don’t have to worry about the current value affecting a later thread re-using the connection.