Dear Dave,
intr_queue_drops was rising very quickly, after changing queue_maxlen to 500
(from 50) the problems all went away. What's a reasonable value for that for
a high load box (5000+ interrupts a second) ?
http://devwiki.pfsense.org/Tuning?show_comments=1
they set it to 3000
intel em-cards have a buffer of max. 4096
some more talks & tuning:
http://osdir.com/ml/freebsd.devel.net/2002-10/msg00072.html
http://osdir.com/ml/freebsd.devel.net/2002-10/msg00104.html
Also, while over a 1000 pings have gone through fine now, a minute or two
after I made the change I saw this --
64 bytes from x.x.x.x: icmp_seq=25 ttl=255 time=0.465 ms
ping: sendto: No buffer space available
ping: sendto: No buffer space available
ping: sendto: No buffer space available
ping: sendto: No buffer space available
ping: sendto: No buffer space available
ping: sendto: No buffer space available
ping: sendto: No buffer space available
64 bytes from x.x.x.x: icmp_seq=35 ttl=255 time=0.294 ms
It hasn't happened again (and I hope it doesn't) - but is that something I
should look for?
try todo a netstat -m when you get the buffer errors.
and increase:
net.inet.tcp.sendspace
just to be shure, you have a 1000mbit link?
Kind regards,
Ingo Flaschberger
_______________________________________________
freebsd-net@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-net
To unsubscribe, send any mail to "[EMAIL PROTECTED]"
