received 0 length packet from server, apparent master shutdown:

[Duncan Berriman]

Hi,

I've got a master db and 3 slaves, 2 local which are working fine and one remote which works fine until it catches up with the master and there is nothing to do.

It then starts causing hundreds of tcp connections per minute and logging errors like this.

050519 23:20:34 Slave: connected to master 'user@host:3306',replication resumed in log 'file.nnn' at position 877232856
050519 23:20:34 Slave: received 0 length packet from server, apparent master shutdown:
050519 23:20:34 Slave I/O thread: Failed reading log event, reconnecting to retry, log 'file.nnn' position 877233728

This is repeated 100's of time per minute with the file positions incrementing as changes are processed.

It appears when it runs out of things to do it gets a zero packet, thinks this is an error, drops the connection and reconnects.

I've checked for similar problems, my server ids are all different, the versions are all the same (4.0.17-standard).

It runs fine until it catches the master up and then the problems start.

I had this on another server at the same location and it worked fine and it was ok on this server for a while but then this started happening and I can not solve the problem.

Any ideas?

Thanks
Duncan