what operating system are you running? not all systems support TCP_NODELAY equally. If a system has poor TCP_NODELAY support, i would imagine it could cause this behavior.

At 12:44 PM 01-15-2003, [EMAIL PROTECTED] wrote:
Hello all,

I have a strange problem with tcpserver and the pop3 daemon.

The short story:
When I start the pop3 daemon through tcpserver with the -D flag the daemon
refuses to die when I run svc -d /service/qmail-pop3d. Without the -D, the
pop3 server is extremely slow and is serving only 3-4 messages per second.

The long story:
tcpserver refuses to die using svc -d /service/qmail-pop3d when the startup
script looks like this:


    tcpserver -H -R -D -lmail.yourplana.com 0 pop3 \
    /var/qmail/bin/qmail-popup mail.yourplana.com \
    /home/vpopmail/bin/vchkpw /var/qmail/bin/qmail-pop3d Maildir 2>&1

If I try to start the pop3 daemon again with svc -u /service/qmail-pop3d I
get the well known "tcpserver: fatal: unable to bind: address already used"
error message. To kill the tcpserver I have to run 'lsof | grep tcpserver'
and then kill the pid of the tcpserver -OR- just 'killall tcpserver'. Then I
can start the pop3 daemon through tcpserver again.

When I'm using this startup script I' having no start/stop problems, but the
pop3 daemon is terribly slow and hands out only 3-4 messages to the client
per second:


    tcpserver -H -R -lmail.yourplana.com 0 pop3 \
    /var/qmail/bin/qmail-popup mail.yourplana.com \
    /home/vpopmail/bin/vchkpw /var/qmail/bin/qmail-pop3d Maildir 2>&1

Please note that the only thing changed here is the tcpserver startup flags.
The first startup script has the extra -D and won't die, the second one is
without the -D flag and is terribly slow but dies nicely.

All the -D flag does is telling the tcpserver not to delay sending data, so
I can't imagine why it becomes a problem when stopping the tcpserver.

Any ideas?


Paul Theodoropoulos
The Nicest Misanthrope on the Net

Reply via email to