I'm working on a project involving distcc. The final aim is to set up some sort of compile farm, for the first serious attempts they would be at least 4, LAN connected and dedicated boxes...
At the current state I'm trying to experiment with the settings, specially the "-jN" option on two gentoo boxes (both with distcc 2.18, gcc 3.4.2 and ccache 2.3, single processor). My aim is to use the remote host as much as just possible, so that the client system is able to work normally w/o a great load. I haven't timed the process as I was just trying to figure out how to set everything up and so I did an emerge -u world to get lot's of compilations running so I could target on distcc... The results are that at least nothing crashed, but I can't tell if things where faster or not. What I would like to know is the opinion of the gurus about my settings and whether this same settings would be usable on the system described above. These are my settings: Host "-J10" Client="-j8" CC="ccache distcc gcc" hostlist="192.168.0.3 127.0.0.1" (distccd running on both boxes) My theory was that launching many parallel builds on the client machine distcc would care about catching any thread which wasn't caught by the local gcc, and that being the remote host the first entry the machine would catch as many threads as the network connection could handle to transmit before the local box. Is this correct? I at least did not notice a high load on the client side (running X, KDE, Mozilla...), just a bit slower, but not specially. On the host side, (-j10) I didn't get more than 60% system load, even when top reported onto 6 or 8 distcc processes. This means either that my settings are right, or that your app is so damn kewl that it can even survive a lame noob like me ;) -- R3G4RDZ __ distcc mailing list http://distcc.samba.org/ To unsubscribe or change options: http://lists.samba.org/mailman/listinfo/distcc