MySQL replication silently failing

We're using standard MySQL as supplied in the RHEL 6 repos. Every so often on an unstable link (the VPN to our office from the DC) i've noticed that although MySQL thinks replication is fine, it's just stalled.

Both slave SQL and IO are running, Seconds Behind Master is zero, no errors at all. To get it back I have to 'stop slave;start slave;' and it bursts into life.

So the questions are - has anybody else seen this? And if so, how did you fix it? The only reason I saw this one quickly is that we've got a cron job on the master updating a date on a table which is what Nagios uses on the slave to determine if it's up to date or not.