On 02/06/11 10:07, William Bakken wrote:
The second log line on the last email shows that we are logging the
IP, in that case, google.

Our resolv.conf points to several DNS servers for our ISP, which all
seem to always have an immediate answer when we do an nslookup
manually.

Is squid caching negative dns queries?  Is there a way to flush these?

Here is the text I am referring to as the last email.

Upgraded squid to 3.1.12

DNS is working from that machine, we are able to resolve both
www.carfax.com and carfax.com, and have even restarted squid after
looking those up.

 From the squid log:
TCP_MISS/503 4367 GET http://www.carfax.com/ - DIRECT/www.carfax.com text/html
(which takes forever)

differs from something like google:
TCP_MISS/200 15049 GET http://www.google.com/ - DIRECT/74.125.93.147 text/html

Nevermind. The default log in the version still outputs FQDN if known. It is fixed to show correct IP in the next series.


Any advice?  Our ISP did switch DNS servers recently (we switched at
the same time as the squid update) but we aren't seeing this behavior
with other sites.


I looked a bit further and found that the NS hanging occurs reliably when it is asked for AAAA records. Such as what Squid-3.1 asks for.

It is pretty clear that the site DNS is screwed bigtime.

Amos
--
Please be using
  Current Stable Squid 2.7.STABLE9 or 3.1.12
  Beta testers wanted for 3.2.0.8 and 3.1.12.2

Reply via email to