severity 419833 normal
tag 419833 moreinfo
thanks
On Wed, Apr 18, 2007 at 08:29:08AM +0200, Rudy Gevaert wrote:
> Sometimes when logrotate rotates our bind9 log files, it doesn't start
> bind9 again. Altought it stops it. My bind9 logrotate file:
>
> /var/log/named/named.log {
> daily
> missingok
> rotate 7
> compress
> delaycompress
> notifempty
> create 644 dnsadm arcs
> postrotate
> if [ -f /var/run/bind/run/named.pid ]; then
> /etc/init.d/bind9 restart > /dev/null
> fi
> endscript
> }
>
> Running /etc/init.d/bind9 restart myself works.
That script is one you've written yourself? The logfile you're generating
and the logrotate script you give there aren't standard.
Might I suggest you use two invocations of "rndc querylog" instead of
restarting the server? Might I suggest you redirect the output of
"/etc/init.d/bind9 restart" to somewhere other than /dev/null, to try to
find out whether it is giving an error?
On a busy site, bind9 may take more than 2 seconds to save its state and
release its UDP sockets. (Have a look at /etc/init.d/bind9 to see how it
implements the restart function.)
This is very unlikely to be a fault in logrotate.
--
Paul Martin <[EMAIL PROTECTED]>
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
