On Tue, Sep 14, 2010 at 01:12:00PM +0200, Gareth de Vaux wrote: > On Tue 2010-09-14 (04:03), Jeremy Chadwick wrote: > > You're absolutely certain these are all in FIN_WAIT_2 state and not > > TIME_WAIT? > > Yup, > > $ netstat -an | grep FIN_WAIT_1 | wc -l > 57 > $ netstat -an | grep FIN_WAIT_2 | wc -l > 431 > $ netstat -an | grep TIME_WAIT | wc -l > 17
Regarding net.inet.tcp.finwait2_timeout=15000 -- you don't see any improvement at all? That's a bit strange. There's probably something about ipfw I'm not getting. Adding luigi@ to the CC list, since I think he's familiar with ipfw internals. FWIW, I'd recommend keeping an appended log of output from something like this script: while true do echo "`date` -- `netstat -n | grep -c FIN_WAIT_2`" >> /some/log sleep 5 done Just make sure the sleep interval is lower than the FIN timeout. Another thing to try: setting net.inet.tcp.fast_finwait2_recycle=1 and see if this makes a difference. I'm not familiar with this setting myself, but some Google results indicate some people do use it. The sysctl description is a bit vague too. -- | Jeremy Chadwick j...@parodius.com | | Parodius Networking http://www.parodius.com/ | | UNIX Systems Administrator Mountain View, CA, USA | | Making life hard for others since 1977. PGP: 4BD6C0CB | _______________________________________________ freebsd-stable@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-stable To unsubscribe, send any mail to "freebsd-stable-unsubscr...@freebsd.org"