Slave Server Slow in updating

Barry Margolin barmar at genuity.net
Fri Jan 25 22:52:00 UTC 2002


In article <a2s65m$q4q at pub3.rc.vix.com>,
Sam Pointer  <sam.pointer at hpdsoftware.com> wrote:
>
>I am running BIND 9.1.3 on Redhat 2.4.7-10 and have a master with a single
>slave. There are no firewall complications. My slave server has the
>following SOA record:
>
>$TTL 0
>mydomain.com.	IN	SOA	ns1.mydomain.com. root.mydomain.com. (
>2002012302 3m 3m 1w 0 )
>
>which I know are agressive values, but I'm testing and want my changes to
>propigate fast.

Zero TTLs have been known to confuse some servers, I suggest you use $TTL 1
instead.  But this is unrelated to your query.

>In my named.conf of my slave server I have the entry:
>
>zone "mydomain.com" in {
>	type slave;
>	masters { xxx.xxx.xxx.xxx; };  <- address of ns1.mydomain.com.
>
>I would suspect then that if I update the serial number on the master then
>the changes would take 3mins to move to the slave. Instead they seem to much
>much longer. Am I missing something fundemental?

I don't know if this is still true in BIND 9, but in BIND 4 and 8 it only
checks for zones whose Refresh time has passed every 15 minutes.

If you have the slave server in the NS records of the zone, the master
should send a NOTIFY message to the slave, which will cause it to check
immediately (it actually waits a random amount of time, so that all the
slave servers don't bombard the master simultaneously).

-- 
Barry Margolin, barmar at genuity.net
Genuity, Woburn, MA
*** DON'T SEND TECHNICAL QUESTIONS DIRECTLY TO ME, post them to newsgroups.
Please DON'T copy followups to me -- I'll assume it wasn't posted to the group.


More information about the bind-users mailing list