Two transactions read one row, both update it, and one update vanishes (or two threads update the same record). What prevents it?

Question

Tests whether you name the lost-update problem by name and reach for either a higher isolation level/row locking (SELECT ... FOR UPDATE) or optimistic locking with a version column, rather than application-level synchronization, which doesn't protect DB state across instances.

Two transactions read one row, both update it, and one update vanishes (or two threads update the same record). What prevents it?

Next Step