We are using Java EE and are making an application where in the worst case, a lot of message queue messages will come form the same User.
Therefore, we are looking on Pessimistic,
SELECT FOR UPDATE style locking. Which in theory and first tests solves our problems.
Yet, we are afraid of deadlocks. Not the classical ones: User X locks A, User Y locks But more of scenarios like: system crash, network problems,... and so on. Database system locking up, for no know reason. We will use modern databases like: Oracle, MS SQL and PostgreSQL.
What we would like to know is pessimistic locking used in production and what practical problems to expect?
Thanks in advance!