Kinkie schrieb: Hi Kinkie, > you run two instances, say one on port 80 and one on port 81. > Then: > iptables -t nat -A PREROUTING -m conntrack --ctstate > RELATED,ESTABLISHED -j ACCEPT > iptables -t nat -A PREROUTING -p tcp -m statistic --mode random > --probability 0.5 \ > -m tcp --dport 80 -j REDIRECT --to-ports 81 > > For more instances, use more ports, and replicate the second line as > many times as needed (also remember to change the probability). Interesting idea. Since there are lots of free CPU-cycles to burn I will look into this. > squid 2.7 supports a '-I' command-line option, which will make it > accept requests on stdin instead of a socket. I recall seeing (but > unfortunately can't find) a PERL script which performed the network > part of accepting a request, and then maintained a farm of squid > servers to which it passed requets this way. I'm not a big fan of constructs like this one. Also I don't have the time to build this "server" myself. > The iptables method is probably more efficient, as the load-balancing > is done directly in the kernel. This definitely looks promising. If my time allows it I'll give it a try. Thanks, Markus
