named start-up behavior

Barry Margolin barmar at alum.mit.edu
Fri Aug 27 01:31:26 UTC 2010


In article <mailman.446.1282839053.15649.bind-users at lists.isc.org>,
 "Gordon A. Lang" <glang at goalex.com> wrote:

> I have not been able to locate documentation defining the named
> start-up behavior.  I am particularly interested in zone loading
> for slave zones.  Is this information available online?
> 
> For example, what if none of the listed masters are reachable at
> the time of start-up?  How frequently and for how long is it
> retried?  Can this be tweaked?  Can the data on disk (if present)
> be used as an interim better than nothing source?

The SOA Retry, and Expire timers control how frequently BIND will retry 
failing zone transfers and how long they be retried.

> 
> If my memory isn't failing me, I thought the old BIND code used
> disk data for slaves before zone transfers -- can anyone confirm
> or refute this?  Can anyone explain why a slave should never load
> from disk (in the event of a zone transfer failure)?

That's exactly what happens.  What do you think the file is for, if not 
to load it at startup?

BIND uses the file's timestamp to schedule the refresh timer.  So if the 
SOA Refresh time is 6 hours, and the file is 1 hour old when BIND starts 
up, it will schedule a zone transfer for 5 hours later.

-- 
Barry Margolin, barmar at alum.mit.edu
Arlington, MA
*** PLEASE don't copy me on replies, I'll read them in the group ***



More information about the bind-users mailing list