Re: CLOSING not closing?

Eric.Schenk@dna.lth.se
Mon, 21 Apr 1997 17:40:40 +0200


"Todd T.Fries" <toddf@acm.org> writes:
>this particular connection has been CLOSING for two days now, and somehow
>sendmail died about that time, and nothing can bind to local port 25 because
>it is 'in use', therefore I cannot even run sendmail.
>
>Is there a userland way to 'kill' this socket or is the only
>solution to reboot?

Reboot is the only way to get rid of it. Clearly the fact that the
socket does not go away is a bug. I'd ask what netstat -not shows,
but I know it won't show anything interesting because the kernel
output in /proc/net/tcp was broken with regard to timers until a
week or two ago.

Unless David or someone else has caught this bug without my noticing it,
we still have the problem with stuck sockets in the latest kernels.
Reports from a recent kernel together with "netstat -not" samples are
of interest. Anyway, hopefully the code cleanup David and I are
currently doing will reveal the problem. If not we will hunt it down
specifically.

-- 
Eric Schenk                               www: http://www.dna.lth.se/~erics
Dept. of Comp. Sci., Lund University          email: Eric.Schenk@dna.lth.se
Box 118, S-221 00 LUND, Sweden   fax: +46-46 13 10 21  ph: +46-46 222 96 38