Sample server fails to bind to IPv4 address

[Michael O'Rourke]

Hi all,

I have 3 linux servers (SuSE 8.0, 9.1 and 10.0) with three different
versions of SASL (2.1.17, 2.1.18 and CVS from 16 March). On all three
machines, the sample server fails to bind to the IPv4 address:

# ./server -p 3500
trying 10, 1, 6
trying 2, 1, 6
bind: Address already in use

Port 3500 is not used by anything else (according to netstat -na).

When I am testing on the local machine, there is no problem as the
client and server will use the IPv6 address it decided to bind to.

My problem is with getting a Solaris 9 machine to authenticate to the
SuSE 10.0 machine (and vice-versa) and I can not use IPv6 between
these two machines.

On the SuSE 10.0 machine, SASL (the CVS version) is compiled with these options:
./configure \
  --enable-static \
  --enable-shared \
  --enable-java \
  --enable-checkapop \
  --enable-cram \
  --enable-digest \
  --enable-otp \
  --enable-plain \
  --enable-login \
  --with-bdb-libdir=/usr/local/BerkeleyDB.4.4/include \
  --with-bdb-incdir=/usr/local/BerkeleyDB.4.4/lib \
  --with-saslauthd \
  --with-openssl=/usr/local/ssl \
  --enable-ldapdb \
  --with-ldap=/usr/local

gcc is version 4.0 (but the other two SuSE machines are the versions
that came with them; i.e. 2.95 or 3.x)
OpenSSL: 0.9.8a
OpenLDAP: 2.3.20

As I do not have a linux machine that the sampe server actually works
on (on IPv4 anyway), it is hard to troubleshoot. I compiled the sample
programs with the -ggdb turned on, but a gdb run does not provide any
useful info.

Does anyone have any ideas?

Thanks, Mike.