On Tue, 2003-11-18 at 19:40, Herbert Poetzl wrote: > On Tue, Nov 18, 2003 at 07:15:11PM +0100, Dariusz Rubinkiewicz wrote: > > -----BEGIN PGP SIGNED MESSAGE----- > > Hash: SHA1 > > > > Dnia Tue, Nov 18, 2003 at 05:08:26PM +0100, Herbert Poetzl napisa(a): > > > okay, this is a trace of the 'forking' parent > > > probably because it daemonizes() ... so we either > > > need a run with '-debug' or whatever keeps it from > > > forking, or strace -fF -s 4096 > > > > > > strace -o log2.txt -fF -s 4096 /etc/init.d/courier-pop start > > > > http://debesciak.net/log2.txt > > well, it seems that this piece of code tries to > attach to an ipv6 address: > > 5787 bind(5, {sa_family=AF_INET6, sin6_port=htons(110), inet_pton(AF_INET6, > "::ffff:62.181.4.38", &sin6_addr), sin6_flowinfo=0, sin6_scope_id=0}, 28) = -1 > EADDRINUSE (Address already in use) > > which is not supported (at least not the virtualization) > by vserver, so stopping the courier-pop from doing so > might help a lot ... >
It's the IPv6 brokeness of vserver that i discussed with Paul Sladen at DebConf 3. courier-imap / courier-pop3 is one of programs, that tries to bind "*" at ipv6 level only and thus also asks for ipv4 addresses. However that way it circumvents your vserver contexts and binds port 143 (or port 110) at all ip's. Or: it simply fails because it can't get them all. Don't use vserver and ipv6 together, it hurts your security. I only use them together on systems that I alone have access to. Regards, Martin List-Petersen martin at list-petersen dot se -- All language designers are arrogant. Goes with the territory... -- Larry Wall
signature.asc
Description: This is a digitally signed message part
