listen-to clusterIP address

paul paul at actionlans.com
Wed Jun 5 09:31:23 UTC 2013


Thanks again. I am using linux fedora17. I have tested changing
interface-interval to i min and that seems to work. I am also looking on
the cluster mailing list to see if I can include named in my cluster
configuration. Paul
On Wed, 2013-06-05 at 10:02 +0100, Phil Mayers wrote:
> Peter Andreev <andreev.peter at gmail.com> wrote:
> 
> >2013/6/5 Phil Mayers <p.mayers at imperial.ac.uk>
> >
> >> On 06/05/2013 07:37 PM, paul wrote:
> >>
> >>> Hi. I have a two node active passive cluster serving webpages. When
> >a
> >>> failover occurs, I have to restart named on the now active node
> >because
> >>>
> >>
> >> You don't have to restart it. "rndc reconfig" will re-check the IPs
> >on the
> >> machine and re-listen.
> >>
> >>
> >This definitely will not work if BIND dropped privileges after start.
> >
> >
> >>
> >>  the cluster Ip was not available when named originally started even
> >>> though I have listen-to the cluster ip listed in my named.conf. Is
> >there
> >>> a way to make named listen-to an ip address that is not yet
> >available?
> >>>
> >>
> >The cimplest way, I think is to configure cluster IP on loopback
> >interfaces
> >and set up routing
> >
> >
> >>
> >> No. This has come up before - the bind listen-on statement is an ACL
> >which
> >> is matched against the list of IPs on the box, not a list of IPs
> >passed to
> >> the bind() syscall. There are various solutions, but "rndc reconfig"
> >is the
> >> right one IMO.
> >>
> >> ______________________________**_________________
> >> Please visit
> >https://lists.isc.org/mailman/**listinfo/bind-users<https://lists.isc.org/mailman/listinfo/bind-users>to
> >unsubscribe from this list
> >>
> >> bind-users mailing list
> >> bind-users at lists.isc.org
> >>
> >https://lists.isc.org/mailman/**listinfo/bind-users<https://lists.isc.org/mailman/listinfo/bind-users>
> >>
> 
> As has been pointed out to me off list, this depends on your os. It definitely *does* work under Linux. But apologies for assuming!




More information about the bind-users mailing list