recursive-clients recommended values
ssb at umn.edu
Mon Jul 16 16:33:50 UTC 2012
One thing that has always been a mystery to me is what the difference is
between the hard and soft limits on recursion - i.e. the default limit on
recursion is 1000 which means that the soft limit is not in effect. When
the limit is reached, the oldest query is always dropped, but without a
soft limit, the most recent query is also dropped. When recursive-clients
is set above 1000, the soft query limit is set to 100 less than the
recursive-clients value and the most recent query is allowed to continue as
long as the hard limit is not exceeded.
I have never found anything in the documentation that describes this
behavior or explains why the soft limit defaults to disabled. It has
always seemed to me that it is more desirable to allow the most recent
query to proceed whenever possible, hopefully allowing named to keep doing
recursion for those sites for which it can while dumping queries for those
which are slower in responding, hopefully allowing some queries to succeed
during a network brownout or loss of connectivity.
On Mon, Jul 16, 2012 at 12:14 AM, blrmaani <blrmaani at gmail.com> wrote:
> On Thursday, July 12, 2012 3:49:27 AM UTC-7, Niall O'Reilly wrote:
> > On 12 Jul 2012, at 03:21, blrmaani wrote:
> > > I searched earlier posts but noticed that people are recommending
> it to just increase it to suppress the errors in log.
> > >
> > > Any pointers on this?
> > If it's set too low for your normal operating circumstances,
> you do need to increase it.
> > I've never needed to do this, as the default values just works
> for me.
> > In abnormal operating circumstances, it's probably neither
> posssible, nor useful to try, to eliminate
> > the log messages. See, for example,
> > Best regards,
> > Niall O'Reilly
> I saw a related post and response in other group.
> I am still trying to figure out the '90 second' part...
> Please visit https://lists.isc.org/mailman/listinfo/bind-users to
> unsubscribe from this list
> bind-users mailing list
> bind-users at lists.isc.org
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the bind-users