SHOBHA HALDONKAR <shobha(at)phildigital(dot)com> writes:
> Suppose if a record is locked with update command and another
> concurrent user tries to lock the same record and the first transaction
> for some reason , eg. m/c hangings is not committed both the
> transactions remains locked for indefinite period . I have checked the
> setting in postgresql.config for deadlock_timeout which is set to 1000
> . Than why doesn't the transaction deatect deadlock after 1000
> millisecond .
Waiting for a client that chooses not to commit is not a deadlock.
regards, tom lane