At 10:38 AM -0400 7/24/01, Daniel Senie wrote:
> We see occasional trouble when a user's connectivity goes south.
> We'll see the TCP session close down, but the popper process
> doesn't close and clean up. Anyone else see this? I'm still trying
> to track down the particulars.
>
> When it happens, users call because they've hung up the phone and
> redialed, and qpopper gives them an error since their mailbox is
> already in use. We try to explain they need to be patient and give
> it a few minutes, but it seems like the sessions should close down
> quicker in the event of the client going away (if TCP knows, why
> doesn't popper?).
It seems Qpopper is not being told that the user disconnected.
Normally, Qpopper either gets a SUGHUP signal, or an I/O error on a
network read or write.
Try using whatever system call trace utility exists for your
platform, such as strace(1) or truss(1). Trace the hung stack, to
see what it is doing.
