In that case, to eliminate a possibility, try recompiling but
include --enable-assert with the configure line. You should compile
with -O2 again too.
Run it without valgrind, and see what happens. :-)
Alasdair
On 7 Oct 2005, at 21:32, Philip Craig wrote:
Actually, not quite.
The situation is:
If I run it -O0 or the normal -O2 via the normal means (command
line or debian init.d script) then it crashes in 3-6 hours.
If I run it under valgrind as either -O2 or -O0, then it stays up
for days!! The first run stayed up 1 day and 1 hour before I
interrupted it. The current one has been up 1.5 days and counting.
So, I have a long term stability workaround it seems. Run it under
valgrind! btw, valgrind reports nothing mysterious if I interrupt
rageircd normally and check valgrind's output. The usual issues
inside sprintf in the libc, and that's it.
So, a great mystery.
Alasdair McWilliam wrote:
Hi
From the lack of communication I assume your server has stabilised
by compiling with -O0 instead of -O2?
Rather odd.... I have a gut feeling this is going to take a
serious amount of debugging. :-(
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
