I forgot to mention a couple of other things.

My infrastructure guys tell me they see no problem with the network link between principal and mirror.

A few times I continously pinged the mirror box (for about an hour at a time) from the principal most of the time the return trip time was less than 1ms but around the time my procedure ran slow I got a couple (no more than 5) return trips around the 25-35ms. Because there was only a couple of slow rtts infrastructure did not think it was an issue.

A few times last week our System Center Operations Manager system complained that databases on the suspect database instance had changed there mirroring state from synchronized to synchronizing. Gilamonster mentioned that the management studio will not show a synchrozing state unless the databases are a lot out of sync - would it show synchronizing if the database where out of sync for more than 10 seconds because sometimes my proc can run for more than 15 seconds.