This is getting a little off-topic, but..
> When it happens again:
>
> killall -HUP named
I don't know, how do you know he's running named on that server? If it is
local, running an strace on the named process to find out what is wrong
might be interesting. I like my processes to not break; just restarting
without finding out what is going on means you will spend time on it later.
You can diagnose the problem as name resolution-related for certain with
careful use of tcpdump and possibly strace. If your issues are
name-resolution related then you need to fix them. The easiest way to do
this is either with local /etc/hosts files or on the name-server end.
Personally I prefer using both for redundancy. I keep my hosts files in
sync with expect scripts and ssh over private interfaces. I keep my DNS
servers working by not trusting ISP dns servers as they usually suck in
multiple ways.
Cheers, TGIF!
P
|