On Thu, 12 Dec 2002, Vincent MATHIEU wrote:
>> If an error occurs with another replica, no reject file is created.
> Then how do you know that an error occurred?
My configuration was bad, and no reject file was created, except for the
first replica
So, was an attempt even made to replicate? What does debugging output
show? Note that tools such as "truss" may be thread-unaware (I got bitten
by this when I found slurpd was using a thread-unsafe function inside a
thread). Fire up GDB with some creative breakpoints, or use "-d".