[squid-users] Extremly high duration value in access.log

Hello squid users,

I'm experiencing some strange timeout problems, that causes pictures not to be loaded from
our backend machines. We are running squid 2.5STABLE7 as a reverse proxy. In front of this
proxy operates an apache-2.0.5x to handle some redirects etc.
Most requests are forwarded to the squid who forwards the request to a backend cluster via
a load balancer.

In our squid access.log we have lines like:

1103204869.900 1 10.1.10.1 TCP_IMS_HIT/304 251 GET http://www.rp-online.de/image1.jpg - NONE/- image/jpeg
1103204870.586 300081 10.1.10.1 TCP_MISS/200 2929 GET http://www.rp-online.de/image1.jpg - NONE/- image/jpeg

The second field is the "duration". But what does a value > 300.000 mean?
Might this be the cause for the spurious timeouts?

The problems only occur, when requesting the pages via reverse proxy. Fetching directly through
the loadbalancer works fine.

Any suggestions?

Thank you all in advance
Stefan
Received on Thu Dec 16 2004 - 07:08:14 MST