[distcc] statistics question

[Martin Pool]

On 14 Jul 2002, Stephan Stapel <stephan.stapel at web.de> wrote:

> I tried using the latest version of distcc and consider it as being really
> great.

Great, I'm glad you like it.

If you could send me some rough performance numbers and information
about what you're using it to build that would be great.  Just at the
level of

  4 x Pentium IV machine running Linux
  building 100,000 lines of C++ in the SuperFoo application
  without distcc: 50 minutes
  with distcc: 20 minutes

> However it currently lacks of features in terms of the analysis of
> the compile process, i.e. the chance to monitor the compile farm and to get
> statistics about the compilation, workload of the machines etc.
> As I didn't manage to inspect the sources yet I would like to know whether
> there are already facilities to do so, i.e. to query workloads of the
> machines, association of machines with certain jobs etc. I know that there's
> no graphical client for this yet but maybe these things exist on the tcp/ip
> level...

You can get some information from the log file, perhaps in conjunction
with a Perl script:

  DISTCC_VERBOSE=1 DISTCC_LOG=$HOME/distcc.log

I know it would be nice if there were clearer statistics and a
graphical monitor.  I'll get around to it when I can.

-- 
Martin