Thanks for help.
The problem is that I can have "some" external IPs and hundreds ports for each IP on the same box.
Up to now, I am using virtual machines for IPs and I route the ip:ports with iptables to the right VM (hundreds ports each). There one squid instance is listening to 128 ports (squid limit). It works very well.
Some customers want more power, so I need to give some VMs more cpus (I have), and run more than one squid process on the same hundreds ports.
Designing loadbalanced configuration with iptables, or frontend/backend with squid for incoming hundreds ports, will get to thousands ports inside.
It is possible I think, but building the configurator is something tiny. Also, configurations can change while in production with squid -k reconfigure for thousands ports.
Of course I can split the ports to separate squid instances, and limit each port traffic.
I need to think well about the solution. Patrick Le 15/07/2014 00:24, babajaga a écrit :
Besides SMP, there is still the "old fashioned" option of multiple instances of squid, in a sandwich config. http://wiki.squid-cache.org/MultipleInstances Besides described port rotation, you can set up 3 squids, for example: one frontend, just doing ACLs and request dispatching (carp), and 2 backends, with real caching. This variant has the advantage avoiding double caching, which might happen in the port rotation alternative. -- View this message in context: http://squid-web-proxy-cache.1019090.n4.nabble.com/Problem-to-set-up-multi-cpu-multi-ports-squid-3-3-12-tp4666906p4666915.html Sent from the Squid - Users mailing list archive at Nabble.com.
