On 26/10/2006 16:24, Jan Johansson wrote:
> Seriously, I am at the point where I might consider paying someone to figure this out.
You probably need a combination of the +all log selector, running Exim
in debug mode in the foreground in a shell, or running strace on the
running Exim daemon.
I'd personally favour the first and then the last.
http://www.exim.org/exim-html-4.63/doc/html/spec_html/ch48.html#SECTlogselector
Add "log_selector = +all -memory" to the top of your config file (it's a
global option) and restart Exim.
Send a message which causes the error, make sure it does cause the error
amongst the noise in the logfile, then remove the option and restart
Exim again. You can then work from one very noisy bit of log for that
message alone.
If that doesn't show up where the problem lies, make a note of the
listening Exim daemon's PID and then do:
strace -fFvp $PID -o /tmp/exim-sympa-trace.$PID
Send a message which causes the error, make sure it does cause the
error, and then CTRL-C out of the strace.
You'll get an absolute mass of debug info with that which should, in
theory, show you which user Exim is running as at the point the error
occurs.
Graeme
