Slave I/O thread dies, fatal error 1236

[matt ryan]

040901 18:36:21  Error reading packet from server: binlog truncated in 
the middle of event (server_errno=1236)
040901 18:36:21  Got fatal error 1236: 'binlog truncated in the middle 
of event' from master when reading data from binary log
040901 18:36:21  Slave I/O thread exiting, read up to log 
'FINANCE-bin.186', position 7517914


2 slaves on one box

both masters have identical config

one slave will not stay running for more than an hour, it seems that it 
will run until it catches up with the master, then it dies.

Any sugestions?  All servers have plenty of free drive space