I get that problem too with the supposedly better Outlook 2002 (XP), and yes I got hundreds of users reporting the problem and asking me to fix it. My response was a politically correct version of "So what? Just ignore it" and eventually the complaints died down. I think it's just Outlook being stupid (as usual). Outlook has so many bugs and stupidities it's unreal; 2002 is a bit better, stopping among other things allowing users to put spaces in email addresses which really bugged me as our smtp server (qmail) took it as two separate email addresses and I got all the resultant bounces, but there are still many things it does wrong or illogically.

Changing the timeouts made no difference for us. I set the timeout to 15 minutes (Outlook 2002 allows this higher setting), but it times out well before 15 minutes anyway, so I don't know what that timeout refers to, and of course the Outlook help is no help.

Increasing the max allowed connections is a no-no for me as I have typically 300 users connected all the time, and 1200 connections is probably about as much as the machine would competently handle.

The problem seems to have improved (on our Solaris 7 server) after I reduced the timeout for dead connections server-side with the following commands:-

ndd -set /dev/tcp tcp_keepalive_interval 120000 # 120 seconds - was 7200000 ndd -set /dev/tcp tcp_ip_abort_interval 120000 # 120 seconds - was 480000 ndd -set /dev/tcp tcp_time_wait_interval 120000 # 120 seconds - was 240000 ncc -set /dev/tcp tcp_conn_req_max_q 1024 # was 128

but I don't know for sure if it was that fix, or the users just aren't reporting it any more. Certainly on my own setup, I see it much less often.

Hope that helps a bit.

Best Regards, Brian

MONZ wrote: