[6376] in Kerberos
Re: What happens when the master node fails?
daemon@ATHENA.MIT.EDU (Mark W. Eichin)
Tue Dec 19 00:11:20 1995
Date: Mon, 18 Dec 1995 23:57:06 -0500
From: "Mark W. Eichin" <eichin@cygnus.com>
To: Sam Tawfik <sam.tawfik@sandiegoca.attgis.com>
Cc: kerberos@MIT.EDU
In-Reply-To: "[6371] in Kerberos"
It's actually pretty straightforward (you're running kprop slaves
anyway, right? just pick one to be the new master, move your CNAME's
and start a new kadmind.) (Oh, and if this is V4, use the CNS code,
not the MIT code, since the MIT code only works for the master having
a *primary* name of "kerberos", whereas the CNS code looks in the
krb.conf to see what machine is listed as the "admin server", giving
far better local control.)
If you're interested in a more automated way of handling it, there are
lots of possibilities depending on what the underlying needs are...
_Mark_ <eichin@cygnus.com>
Cygnus Support
Cygnus Network Security <network-security@cygnus.com>
http://www.cygnus.com/data/cns/
> I was wondering if anyone in this newsgroup have had experience
> with what you you'll have to do if the node with the master
> database should go down and stay down for a few days?
