Odiobill wrote:
Hi, I configured an OpenVZ cluster for my company, where there is a VE with
public IP adress that provides web-accelerator for other VEs that are
running on a different network, with private addresses.
To realize it, I used a configuration similar to this one within my
squid.conf:
-- CUT HERE --<
http_port 80 accel vhost
http_port 3128
acl our_network src 192.168.10.0/24
http_access allow our_network
acl our_domains dstdomain "/etc/squid/sites.conf"
http_access allow our_domains
http_access deny all
cache_peer 192.168.10.100 parent 80 0 no-query originserver login=PASS
name=100
cache_peer_domain 100 1st.example.org 3rd.example.org
cache_peer 192.168.10.101 parent 80 0 no-query originserver login=PASS
name=101
cache_peer_domain 101 2nd.example.org 4th.example.org 5th.example.org
-- CUT HERE --<
/etc/squid/sites.conf lists every hostname referred by cache_peer_domain
directives.
The problem is that we have ten virtual servers to accelerate, and some of
these are providing hosting for 20 or 30 websites (each one).
There is better way to accomplish this, maybe without the need to write down
hostnames both on the external file and squid.conf itself?
Yes there is :-) cache_peer_access.
I use it something roughly like this:
cache_peer .... name=serverA
acl serverA dstdomain "/etc/squid/serverA-domains"
http_access allow serverA
cache_peer_access serverA allow serverA
cache_peer_access serverA deny all
... repeat for each host.
You can then automate the squid.conf generation if you need.
Any alterations to the data files still require a squid
reload/reconfigure to activate.
Amos
--
Please use Squid 2.7.STABLE4 or 3.0.STABLE9
