TTLs' and notifies

[phn at icke-reklam.ipsec.nu]

Steve Foster <fosters at uk.psi.com> wrote:

> At 13:54 15/08/02 GMT, phn at icke-reklam.ipsec.nu wrote:
>>
>>the slave should upon receiving a notify set teh SOA TTL to zero, 
>>which triggers a serial number check and a following zonetransfer.

> or not, if it determines the serial is the same as the one it already has
> and just reloads the zone...
If serial numbers aggree, yes just keep the zone for another "refresh" period.
It's not a matter of "reloading the zone", it's only a matter of 
keeping it ( reloading might be a lengthy process if it's a large zone)

>>
>>Expiration of a zone is something else, it's when a slave hasn't 
>>been able to check serial and starts returning SERVFAIL for
>>queries about that zone.

> so potentially a network issue could cause the domain to expire on my
> secondaries, and not be able to be refreshed until named restarts, or the
> primary sends a notify...

The slaves should check SOA when refresh timer equals zero, 
NOTIFY just set's it to zero. So even if the notify is lost,
querying for s SOA will happen at "refresh" interval.


> i am seeing the domain expire quite often, and also the named-xfer failing
> with premature EOF, so i beleive the 2 could be linked, i have just seen a
> further post from  Danny Mayer suggesting that bind8 on win32 has this
> issue, so i am checking with the customer to find out what he is running.

A likley schenario yes. Make them upgrade to un*x and bind-9

> Cheers guys

> Steve


> Steve Foster
> Senior Systems Administrator
> PSINet Europe
> Work: +44 (1223) 577322
> Mobile: +44 (7720) 425911


-- 
Peter Håkanson         
        IPSec  Sverige      ( At Gothenburg Riverside )
           Sorry about my e-mail address, but i'm trying to keep spam out,
	   remove "icke-reklam" if you feel for mailing me. Thanx.