Question about BIND 8.2.2 on Solaris 7

[BOFH]

For some reason, one slave server that I maintain has stopped looking
for/accepting updates from the master.

The named.conf (see below) is identical to the other slave servers and
the slave in question is in the allow-transfer list on the master.

It seems that following a re-install of BIND, the slave will accept
transfers and updates for about a week (avg 6days, 17max, 1min) before it
simply stops talking to the master.

Reloading the database, restarting the service, killing and then
re-initing the service, -nada- nothing works from that point on.

At this point I am clueless.  The server (a SPARC Ultra-II 300MHz, running
Solaris 7 with the most recent patches as of 11-01-1999) is identical to
the other two slave servers in both hardware and software build.  The
other two never have a glitch, this one however, is rapidly becoming a
migraine.

Any suggestions?

- Ed