We have a SQL Server 2000 Standard database with some tables being replicated (continuous transactional replication) to dozens of SQL Server 2005 Express and MSDE computers.
The step 2 of the replication agent job (Run agent) is configured by default to retry every 1 minute for 10 times if some problem ocurr. Because the client machines get shut down at night (they are POS machines), we changed the number of retries to 5760 (4 days), so replication would not be broken at night and would not need to be restarted manually.
But the problem is that every other day we have at least one machine with broken replication, with this error:
The process could not connect to Subscriber 'POS986'. NOTE: The step was retried the requested number of times (5760) without succeeding. The step failed.
It seems that SQL Server is not respecting the number of retries or the interval between retries as we configured.
PS: I have restarted the replication job after changing the number of retries from 10 to 5760.