[squid-users] more dns problems

Hello List,

last week i reported my dns-caching problems to the list,
unfortunately nobody could help me.
I hope someone can, now.

I am working with squid2.3.STABLE1 on redhat 7.0 linux kernel 2.2.18 .

In my network there are two internal dns servers ( both bind 8 on
solaris 8 ),
i have the problem that squid becomes nearly unusuable as soon as the
first box
is down.
It seems that squid is making a dnslookup for every item (picture,
html-page) that it is trying
to load, and allways trys the nameserver that is configured first in
/etc/resolv.conf
and after the timeout, tries the second one.
But for the next item he starts again with the first, ( which he should
know
it is unreachable for the moment from the lookup before ).

I can see this making a tail -f /var/log/squid/access.log , i can se an
entry appearing,
then nothing for two or three seconds, and then the next entry. But all
are from the
same webpage. Should not squid cache these lookups somewhere? ( that is
what my
last mail was about, too).

Squid is configured to use a parent proxy, but has lots of dns acl´s in
its configuration
file, thats why it´s forced to make dns querys.

I cannot imagine, that squid cannot handle the loss of one of its
nameservers, as this situaion
should happen from time to time in the internet! What are secondarys
for?

Thanks in advance.

timo proescholdt
Received on Wed Nov 21 2001 - 13:49:49 MST