We have a very active (high rate of change) MySQL master / slave. After the slave was offline for a few hours (over a million binlog entries missed) the slave will not start replication again. When you look at the slave status, it shows slave IO & slave SQL running, however, the binlog position never updates.
Doing a "show processlist" on the master shows that the binlog dump starts writing to net, however, after about 1100 seconds it silently dies. However the slave still thinks that it's connected and waiting for the master to send events.
There is nothing in the master or slave error log - the master just silently stops the slave binlog dump with no error message. Doing a fresh dump from the master isn't really an option as the time to dump and restore would leave me back in the same place of being at least two hours behind the master.
Any insight into what's causing the master to silently stop the binlog dump?