[squid-users] Cache Peer Problems

Hi,

We are having problems setting up a cache_peer scenario. What we currently
have is a cache unit that is upstreaming all requests to a cache peer
parent. For redundancy purposes we have added another cache peer parent with
the default option set. Now the problem we are having is that all requests
go to the peer parent with the default option set, even though according to
the manual default will be the last resort. If I then turn off the parent
with default set, the other parent never kicks in and we only get timeouts
at the lowest level.

                Preferred parent Default parent
                  10.10.3.1 10.10.3.3
                        |___________________________|
                                           |
                                     10.10.0.1
                

The error at 10.10.0.1: 2005/12/06 13:07:48| TCP connection to
10.10.3.3/3128 failed

Here are the respective parts from squid.conf on the 10.10.0.1 unit.

always_direct deny all
never_direct allow all
cache_peer 10.10.3.1 parent 3128 0 login=PASS
cache_peer 10.10.3.3 parent 3128 0 default login=PASS

To sum up, all requests on 10.10.0.1 are going to 10.10.3.3 even though it
is set to default. If we kill 10.10.3.3 it still doesn't go to 10.10.3.1
because 10.10.3.3 is set as the default. It never even detects it as a dead
parent.

Any ideas?

Thanks in advance

Ian
Received on Tue Dec 06 2005 - 04:20:09 MST