At 07:24 AM 8/10/2006, Paulo Condutta wrote:
Hi,

After upgrade my server to a HP Proliant 320L run Fedora 5 and Qpopper 4.09
or Qpopper 4.08 and clients run Outlook 2000 ad 2003 and Symantec Corporate
2006, many (no all) users receive error on "get mail".

In maillog write this error:
Aug  7 15:42:34 mail-srv02 popper[5368]: user at 192.168.2.110
(192.168.2.110): -ERR POP EOF or I/O Error

I configure my Qpopper with :
./configure --enable-shy --enable-specialauth --mandir=/usr/man
or
./configure --enable-shy --enable-specialauth --mandir=/usr/man --disable-old-spool-loc
or
./configure --disable-old-spool-loc --enable-shy --enable-specialauth --mandir=/usr/man --enable-spool-dir=/var/spool/mail --enable-temp-drop-dir=/var/spool/mail.drop
 --enable-cache-dir=/var/spool/mail.cache

My basic configuration run Qpopper on Xinetd with flags -s -R -S

Obs. if. disable Symantec 2006 on clients i can receive mail, but this
antivirus using old server run no problem.

Any sugestions ??

Is the Symantec Antivirus proxying, and running its own POP session? I believe this is the case. It would be interesting if you captured the session with the AV disabled on the client, and enabled and compared the messages. I've had more than a bit of trouble over the years with Symantec's methodology of proxying the POP sessions and generally have users bypass Symantec entirely by running alternate port POP (Port 995, with SSL) or just disable the email scanning. Other antivirus vendors have actually used the AV hooks into mail programs (AVG, for example) which makes MUCH more sense than sticking something into the Winsock stack.

Clearly if there's a difference in how things are handled between Symantec and Qpopper, we're interested to learn more, thus the request to have you get traces and compare them. Whether this points to a bug in qpopper, kernel or the Symantec code is not clear from the symptoms, however. You note that you've changed kernels and IP stacks, which may also contribute to heartburn on the part of Symantec. So, there are several variables that all need to be explored. Anything you can discover would be of interest.

Reply via email to