slave server does not reload
Mark Andrews
Mark_Andrews at isc.org
Mon May 2 21:40:18 UTC 2005
> Hi all,
>
> After moving to a new IP block a typo crept into the master zone file
> for one of my domains. The typo was corrected, and the master server
> gave the correct answer.
>
> However, the slave server repeated the error, even after reloading
> from the correctly-functioning master server. Renaming the backup
> file and issuing rndc reload had no effect, the backup file was
> _not_ recreated and the wrong answer was repeated. However, there
> were no error messages, on command line or /var/log/messages.
>
> A reboot of the slave reconstituted the backup zone
> file and fixed the error.
>
> If I made a mistake please let me know, rebooting seems too drastic.
>
> Thanks for reading!
>
> bob prohaska
Sounds like you failed to update the serial number when you
"fixed" the zone.
If you want to force the slave to retransfer a zone use the
"rndc retransfer" command.
With a correctly configured master/slave relationship there
in no need to force the slave to refresh after updating the
master. The master will issue a NOTIFY and the slave will
refresh provided the serial number has been correctly
updated, the master can turn the nameservers name into a
address, etc.
If you want to force the slave to use refresh earlier than
it would normally do then use "rndc refresh" which requires
a zone to be specified or the extended "rndc reload" where
you specify the zone to reload. A plain "rndc reload" does
not cause the slaves to be refreshed as it would create an
enmourmous traffic spike (self DoS) on large (many zones)
nameservers.
Note named also doesn't refresh all slave zones initially for
the same reason. It staggers them using the SOA timer values.
Only non-existant zones get a refresh at startup.
Mark
--
Mark Andrews, ISC
1 Seymour St., Dundas Valley, NSW 2117, Australia
PHONE: +61 2 9871 4742 INTERNET: Mark_Andrews at isc.org
More information about the bind-users
mailing list