hi everybody,
Just a note of a problem I found - to see whether worth reporting this as a bug, or suggest that someone mention it in the manual.
In short: *when replication fails because 'max_packet_size' is too small, that failure can corrupt the relay logs*. So even when you fix the max_packet_size, you get another replication error.
I got into this problem, and I've found a couple of mentions of other people having this problem on Google.
The fix is this: (1) stop slave (make a note of Exec_master_log); (2) set max_packet_size to be larger; (3) do a CHANGE MASTER to re-read logs from the Exec_master_log position; (4) start slave.
Tom.
