Ok... It got stuck... Here is what was in the log:
Mar 21 20:00:28 kermit dovecot[25235]: pop3-login: Disconnected: Inactivity: rip=72.193.85.114, lip=64.69.41.217
When it works right, this is what is there:
Mar 21 20:04:42 kermit dovecot[34276]: pop3-login: Login: user=<don@lizardhill.com>, method=PLAIN, rip=72.193.85.114, lip=64.69.41.217 Mar 21 20:04:42 kermit dovecot[34276]: POP3(don@lizardhill.com): Disconnected: Logged out top=0/0, retr=0/0, del=0/175, size=4992146
Any clues from this? Could it be related to the DirectAdmin pop-before-smtp code? Or maybe stunnel? I haven't updated to the latest RC yet, so this may go away, but it's quite bothersome at the moment.
-----Original Message----- From: dovecot-bounces@dovecot.org [mailto:dovecot-bounces@dovecot.org] On Behalf Of Timo Sirainen Sent: Wednesday, March 21, 2007 6:38 PM To: Don O'Neil Cc: 'Dovecot Mailing List' Subject: Re: [Dovecot] Timeouts connecting via POP3
On Wed, 2007-03-21 at 18:00 -0700, Don O'Neil wrote:
auth default: username_chars: abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ01234567890.-_@& count: 5 passdb: driver: passwd passdb: driver: passwd-file args: /etc/virtual/%d/passwd userdb: driver: passwd userdb: driver: passwd-file args: /etc/virtual/%d/passwd
Where does passwd find the users? Are you using nss_ldap or something? Maybe the hangs are related to it? With newer Dovecot versions you can use blocking=yes setting, which does the passwd lookups in separate dovecot-auth worker processes, and if they use up all the processes it logs an error and returns "Internal error" to client instead of hanging.
Also if you have more virtual users than NSS users, it's faster to put the passwd-file passdb/userdb before the passwd, so that that the passwd-file lookups are done first.