POP3 service aborting

[cswihart]

I have a strange problem with POP3.

I have just begun testing the scalix pop3 service with our user accounts. However I'm having trouble with the POP3 service continually aborting. I can usually complete a send/receive twice with any client (outlook, thunderbird, express) before the service fails. I've checked the fatal log and increased the POP3 service logging and I do not see any indication of a problem. The strange thing is if I enable the UAL_POP3_TRACE=DETAIL in my general.cfg the POP3 service runs fine without aborting.

So what's the difference with POP3 when logging is enabled?

-Chris

[florian]

what version of scalix is this? (rpm -qa | grep scalix)

Florian.

[cswihart]

Sorry, I should have added that to my posting..It's version 11.0.1 for SLES 9.

-Chris

[florian]

right - we did have a pop server crash issue in scalix 11.0.0, that's why i was asking.

we have seen some sles9-specific crash reports; these are tracked as http://bugzilla.scalix.com/show_bug.cgi?id=14709 - however, we can't reproduce them and have no idea what's causing them. can you check if this describes your symptoms?

If not, is there any further information in your fatal file and/or in omshowlog output w/rgds to the crash?

Florian.

[cswihart]

Unfortunately I'm not seeing the same events being logged as whats described in the bug. Omshowlog and the fatal file do not seem to indicate any error at the time the crash occurs.

-Chris

[cswihart]

I think I've narrowed it down a bit. It appears this may be some kind of authentication issue. I'm authenticating all my users against Novell eDirectory. Each time I perform a send receive the user authenticates against the directory. With pop3 accounts this appears to be creating an issue. I can authenticate once without error. However when I go to authenticate a second time Novell logs a TLS accept failure and POP3 aborts. With the UAL_POP3_TRACE=DETAIL enabled this doesn't happen. Could this be a timing issue?
Perhaps the logging is slowing the process down enough to allow the authentication to complete properly?

Here is /var/opt/scalix/xx/s/sys/pam.d/pop3 in case you need to look at it.

Code: Select all

auth sufficient om_ldap
auth sufficient om_auth
auth required pam_deny
account required om_auth
password required om_auth
session required om_auth

This only affects the POP3 service. SWA and Scalix Outlook Connector clients work fine.

-Chris

[ksanders99]

We're seeing the same thing here, fwiw. We're using Scalix 11.1 and bouncing authentication off of Edirectory as well.

Our POP3 daemon also seems to work ok if we leave the UAL_POP3_TRACE switch turned on, but aborts upon connection if we turn it off.

If I do an omshowlog and grep for POP3, I do get a message stating:

ERROR POP3 interface(POP3 Process ) 07.22.07 20:17:51
[OM 29260] PAM unable to dlopen(/opt/scalix/lib/security/om_ldap.so)

ERROR POP3 interface(POP3 Process ) 07.22.07 20:17:51
[OM 29260] PAM [dlerror: libldap.so.199: failed to map segment from shared object: Cannot allocate memory]

ERROR POP3 interface(POP3 Process ) 07.22.07 20:17:51
[OM 29260] PAM unable to open static handler /opt/scalix/lib/security/om_ldap.so

ERROR POP3 interface(POP3 Process ) 07.22.07 20:17:51
[OM 29260] PAM adding faulty module: /opt/scalix/lib/security/om_ldap.so