On 25/09/2015 8:26 a.m., Alex Rousskov wrote: > On 09/24/2015 02:10 PM, Ahmad Alzaeem wrote: > >> If I use 2k ips with 2 worker , squid works ok If I use 10kbports without SMP , squid is ok >> With 10K + 2 workers , we have reg timeout > > The bigger (workers * ports) product is, the more likely you are to run > out of the UDS buffer space because unpatched Squid workers request > sharing of all http_ports at once. > > >> error: "net.local.dgram.recvspace" is an unknown key > > Sorry, I do not know what that option is called in your environment. > > >> if process # 1 , I give it ports 3K If process # 2 , I give it 3 >> K And so on ....will that success ?? > > I am not sure, but I suspect that you will get different workers > listening on different ports, without sharing. It is not a configuration > SMP Squid was designed for, and workers will still send UDS requests to > share their ports (a worker does not know whether other workers are > using its port). It does not hurt to try. > I would just add that if you are able to do this then you should also be able to use a multi-tenant design to scale your Squid horizontally. PS. Since you are obviously building a custom Squid to get past the 128 listening sockets limit anyway. Please do your building with the latest 3.5 series release. The -n option in current 3.5 will let you do multi-tenant easily. Amos _______________________________________________ squid-users mailing list squid-users@xxxxxxxxxxxxxxxxxxxxx http://lists.squid-cache.org/listinfo/squid-users
