[quote user="dkocmoud"]The -Z 1024 commandline option must be used every time, not just once.  [/quote]

Yes, I know. The shortcut I usually use (in fact it's in the start-up folder, so i won't have to do it manually) has this parameter. However, because I saw something that I did not see early on, I dropped it here. The parameter suppresses the symptoms, it is not solving the problem. Maybe someone with more knowledge of the subject finds the info is usefull.

And yes, I know the problem is not a Pegasus-bug. I've reported the bug also by the company whose software I think is tot blame.

However, no hard feelings here. 

Since about two weeks the first time Pegasus checks a mailbox, there is an error saying:

 Pop 3 Network or protocol error

 ........

Trace showing end of transaction:

5: Connect failed, 6. 

Subsequent mail-checkings  are fine.  This is happening at home and at work, both XPpro--systems. The only two changes I can remember are the M$-patchday, and the update from the Norman-AV software thats running on both systems.

Has anyone a clue?

What this indicates is that the low-level networking layer in Windows has reported a failure during connection, and the error value it has returned to the program is "6". There are two important points here:

1:   Pegasus Mail is simply reporting an error that has occurred at a lower level in the operating system - it is not an error in Pegasus Mail itself.

2:   Error 6 is invalid - it does not appear in the list of possible error returns that the Windows Sockets "connect" function can return as a valid error.

It is safe to conclude from this that something operating a level below Pegasus Mail in your operating system is causing the problem, and that it's doing it quite badly (by not even returning a valid error to which I can respond). If you have something that intercepts TCP/IP traffic, such as a firewall, a POP-based antivirus scanner, or a VPN driver, then that would be where I would be starting to look for this problem.

Cheers!

-- David --

Do you have Norton configured to scan incoming POP3 email messages for viruses?  If so, it may be delaying the POP3 connection or returning an invalid command back to Pegasus Mail the first time.

To better figure this out, please enable the session logging feature in Pegasus Mail to obtain a session log of the SMTP transaction.  To do this, click on Tools -> Internet Options, then enable the checkbox at the bottom of the dialog to "Create Internet session logs" and click OK.  Now exit and restart Pegasus Mail in the manner that usually causes the error you're seeing (i.e. may need to restart the PC).  Once you get the error, go back into the Internet Options dialog and disable the session logging feature.  Now check your mailbox directory (the location is noted within the Info text window within the About -> Pegasus Mail... menu option) for TCPxxxx.WPM files (there should hopefully only be one of these, unless you've used this option in the past).  Look through this file for any lines that contain your password and remove the password and save the file.  Then send the file to us (or post it) for analysis.

David Harris wrote:

It

is safe to conclude from this that something operating a level below

Pegasus Mail in your operating system is causing the problem, and that

it's doing it quite badly (by not even returning a valid error to which

I can respond). If you have something that intercepts TCP/IP traffic,

such as a firewall, a POP-based antivirus scanner, or a VPN driver,

then that would be where I would be starting to look for this problem.

End  of quote

 The virus scanner is my first candidate. However, nor the TB-client my wife is using, nor the OE-clients my colleagues are using seems to have this problem. And because I know nobody using the combi Pegasus Mail and Norman virus control, I hoped someone on this forum recognised the problem.

[quote user="dkocmoud"]

Do you have Norton configured to scan incoming POP3 email messages for viruses?  If so, it may be delaying the POP3 connection or returning an invalid command back to Pegasus Mail the first time.

[/quote]

My AV-software is Norman  Virus control. I don't think it's a delay problem, the error appears within seconds.

 [quote user="dkocmoud"]

To better figure this out, please enable the session logging feature in Pegasus Mail to obtain a session log of the SMTP transaction.  To do this, click on Tools -> Internet Options, then enable the checkbox at the bottom of the dialog to "Create Internet session logs" and click OK.  Now exit and restart Pegasus Mail in the manner that usually causes the error you're seeing (i.e. may need to restart the PC).  Once you get the error, go back into the Internet Options dialog and disable the session logging feature.  Now check your mailbox directory (the location is noted within the Info text window within the About -> Pegasus Mail... menu option) for TCPxxxx.WPM files (there should hopefully only be one of these, unless you've used this option in the past).  Look through this file for any lines that contain your password and remove the password and save the file.  Then send the file to us (or post it) for analysis.

[/quote]

 I've done that. By the way, when closing pegasus, the sessions logs are disabled by default, so starting in offline mode and enable session logs has to be done.

This is what the logs shows:
 

 First mailcheck, which goes wrong:


--- Tue, 22 May 2007 07:50:38 ---
Connect to 'mail.noordgouw.nl' port 110, timeout 125.
07:50:38.093 5: Connect failed, 6.


Second, normal, mailcheck:



--- Tue, 22 May 2007 07:50:43 ---
Connect to 'mail.noordgouw.nl' port 110, timeout 125.
07:50:43.359 [*] Connection established to 85.17.3.140
07:50:43.375 >> 0045 +OK POP3 beta [cppop 20.0] at [85.17.3.140]\0D\0A

Snipped here, the rest is like you expect to see.

 Anyone?

 Edit:

I have just made a telnet connection with my pop-box. No glitch there.

What happens if you disable Norton's POP3 checker before starting Pegasus Mail and issuing the first POP3 check?

Does this happen only after a reboot or after each time that Pegasus Mail is started?

BTW, did you try the "-Z 1024" commandline option that was suggested in a similar post?  This option tells Pegasus Mail to use "blocking" sockets instead of "non-blocking" sockets.  Some low-level applications and custom winsocks only work properly with "blocking" sockets.  This commandline option is a "compatibility" option for broken winsocks.

[quote user="dkocmoud"]

BTW, did you try the "-Z 1024" commandline option that was suggested in a similar post?  This option tells Pegasus Mail to use "blocking" sockets instead of "non-blocking" sockets.  Some low-level applications and custom winsocks only work properly with "blocking" sockets.  This commandline option is a "compatibility" option for broken winsocks.

[/quote]

Yes, I found the z -1024 option in the other thread today and tried it. With this option the error seems to be vanished :-).

And to yoou're earlier posting, the problem occurred after every restart of Pegasus, not just after a reboot.

Is http://www.cexx.org/lspfix.htm of interest?

[quote user="magiric"]Is http://www.cexx.org/lspfix.htm of interest?[/quote]

Interesting - that's definitely a little utility to file away in my memory somewhere. Thanks for sharing this with us!

Cheers!

-- David --

Since starting up Pegasus with the -Z 1024 command, the error is gone. However, due to a restart using a shortcut without the parameter the error showed up once more. But just before the error showed, the statusbar showed the message "Downloading message 1 of 1" (or something equivalent). Something I hadn't seen before, normally the error is first.

After dismissing the error-dialog en recheck for new mail, the promised message was downloaded as usual.

For the record, the error  occurs also when sending messages, but not when using IMAP. Just the first POP or SMTP-action triggers the error. 

The -Z 1024 commandline option must be used every time, not just once.  You'll continue to get that error on startup any time the option is not used.  This is due to faulty winsock implementations that don't properly support non-blocking sockets.