Resolv.conf not querying correctly?

[Kevin Darcy]

This is not the behavior I'm seeing with my Solaris 7 workstation. I see one
query to a non-working address, then 5 seconds later a query to the second
nameserver in the list. How are you generating the queries, i.e. nslookup or
just a regular TCP/IP utility like telnet (which is what I used)? How are you
verifying the timings and the query destinations (I used snoop)?


- Kevin

Bjarte Nilsen wrote:

> Hi,
>
> I thought that if you had a /etc/resolv.conf like this:
>
> nameserver 193.1.12.1
> nameserver 127.1.22.10
> nameserver 193.212.1.10
>
> it would first query #1 for 5 sec. then go on to #2 for 5. sec. and last #3
> for 5 sec - before it would go back to #1, double the timeout, and start
> over.
>
> To me it looks like it's querying #1 for 5 sec, then #1 for 10 sec, then #1
> for 20 sec, and last #1 for 40 sec - before it does the same with #2 and
> last #3.
>
> So, if my first nameserver goes down, my clients have a DNS-resolv hang for
> about 5+10-20+40= 75 sec. for every time it needs to desolve a name.
>
> I have noticed this with both Solaris 2.6, 2.5.1, Irix 6.5.1, 6.5.3, HP-Ux
> 10.20, LInux RH 6.0, 6.0.1.
>
> I can implement a workaround in Solaris by setting "options retry:1". But
> this doesn't help me for the other OS's.
>
> Is this normal behaviour?
>
> /Bjarte Nilsen