🚀 go-pugleaf

In <rossc.717550599@extro.ucc.su.OZ.AU> rossc@extro.ucc.su.OZ.AU (Ross Cartlidge) writes:

| Does anyone know a better way of fixing this up -or
| avoiding this in the future?

The idea way is not to change addresses of nameservers, or at least
not until a much much better version of bind comes along.   (You can
delete something as a nameserver, wait, change its address, then make
it a nameserver again, as long as you make sure to totally delete it,
and probably not even run bind in the interim, and take all the
appropriate steps to make sure all secondaries, etc, are updated).

But, more practically perhaps, you really need to have the nameserver
which is having the address expunged not be a secondary for anything
at all, so its not likely to get corrupted, then change the SOA of all
of its zone files frequently (set the refresh period low, and change
the SOA's every refresh period or so).  Keep checking to make sure
your server isn't corrupted, restart it with a change of all SOA's if
its found to have the old address back again somehow.

Check all secondary servers of yours, including unofficial ones,
and all secondaries of any domains that you're also a secondary of
probably, once they're all clear of the old address, and all the servers
you are normally a secondary of are also clear, then wait a bit longer,
and check again, if everything is still clear, you can probably go
back to normal.

Things will clear out quicker if you set the TTL on the old address way
down (perhaps even 0) for a couple of refresh periods before it is
deleted (with an SOA increment of course).

kre