Matt Carey [cvstealth2...@yahoo.com] wrote: > I'm trying to track down the source of what is causing output errors on vlan > interfaces for 2 separate physical systems. ?For example when looking at > netstat between 2 different runs the values are always incrementing: > > # > netstat -s -f inet -I vlan800 && echo && sleep 5 && netstat -s -f inet -I > vlan800? > Name ? ?Mtu ? Network ? ? Address ? ? ? ? ? ? ?Ipkts Ierrs ? ?Opkts > Oerrs Colls > vlan800 1500 ?<Link> ? ? ?00:1c:23:e1:cf:48 187689428 ? ? 0 > 148043392 262767 ? ? 0 >
This output is mostly unreadable. Maybe the yahoo mailer is the problem. > The same behavior is > mimicked on both systems as the counters start incrementing when failing over > the carp interfaces between the peers. Another oddity that is the physical > interfaces show no output errors just input errors: Take a look at /usr/src/sys/net/if_vlan.c, there are three places where it increases if_oerrors. You could try to sprinkle unique printfs next to each if_oerrors++ to see which one is getting triggered. Each one happens for a different reason, parent interface not UP/RUNNING, parent interface doesn't handle its own tagging and if_vlan can't allocate mbuf, etc.
