On torsdagen den 16 december 2010, you stated the following: > after installing a new debian squeeze system, prayer fails to start with > [...] > Error was: [os_bind_inet_socket()] bind() failed: Address already in use > Aborted > > a deep inspection shows that the problem is that no IPv6 Routers are (in > my environment) present.
Actually, the bug is that Prayer tries to bind IPv4 and IPv6 sockets separately without setting the IPV6_V6ONLY flag. That fails because when the sysctl net.ipv6.bindv6only is set to 0 (after a long debate, the default value on Debian), IPv6 sockets listening on [::] will also handle IPv4 packets (with the remote address looking like ::ffff:1.2.3.4). So the problem occurs on dual-stack systems whether IPv6 actually works or not, but not on systems without any IPv6 support at all. But thanks for reporting anyway. I should have noticed it myself but for some reason never really tried starting Prayer with net.ipv6.bindv6only=0. To work around the problem one can set net.ipv6.bindv6only=1 in /etc/sysctl.conf (will affect all software on the system) or set use_http_port to 0.0.0.0:port or :::port as you suggest. -- Magnus Holmgren holmg...@debian.org Debian Developer
signature.asc
Description: This is a digitally signed message part.