Amos Jeffries wrote:
Your cache_peer config REQUIRES squid to lookup the IP of the peer on startup. If that fails it ignores the peer for sanity, and dies. Place the IP of the peer directly into your squid.conf.
You were right, but not for the obvious reason. The peer is a multi-homed host. To avoid the DNS lookups, I had copied the four entries for the peer into my /etc/hosts file many years ago. Unfortunately, either some of the addresses had changed, or I copied them wrong. Two of the addresses were correct, but two of them that I had listed as 129.149.x.x were really 129.146.x.x. I can't tell you how many times I looked at the list and didn't see the difference. The two correct addresses seemed to be good enough for the Netscape proxy server to figure out, but it looks like squid was picking a random address and getting one of the bad ones, and it stopped there. Anyway, correcting the addresses caused everything to work! In summary, here's the changes I made to the default config file: # diff etc/squid.conf.default etc/squid.conf 589a590 > acl localnet src 192.168.1.0/255.255.255.0 633a635 > http_access allow localnet 939c941 < http_port 3128 --- > http_port 8180 1500a1503 > cache_peer webcache.sfbay.sun.com parent 8080 7 no-query 2974a2978 > cache_mgr shannon 3017a3022 > cache_effective_group nobody 3033a3039 > visible_hostname nissan.home.sfbay.sun.com 4071a4078 > never_direct allow all 4184a4192,4193 > # copied from /etc/resolv.conf... > dns_nameservers 129.146.11.51 129.145.155.226 129.147.62.34 4219a4229 > dns_testnames localhost Thanks for your help!
