Steve Woodford wrote:
> Looking at the source for wm(4), it appear that models before i82543 do 
> not support the 1504 byte packets required for VLAN tagging. This is why 
> the vlan(4) reports an MTU of 1496. If you don't tweak the MTU to 1504, 
> does everything work fine? If not, perhaps you are blocking ICMP, which 
> will break path MTU discovery.

   Okay.  Interestingly, tho, the fxp reports 1500, while vlan reports
1500.  Why does the wm's vlan interface report 1496, unless wm is
set to 1504?  Not real important, but an odd difference, IMHO.

   I don't recall.  I was having problems with things when the vlan's
were 1496.  Anytime large packets tried to flow, things would get
confused.  I'd assumed this was an issue of other hosts on then
network using 1500, while this host was using 1496.  And I think
I'm allowing all ICMP on internal networks.

> What does the kernel dmesg say about the wm(4) card at attach time?

   Good question.  Sorry I forgot to put that in the first message.

wm0 at pci1 dev 6 function 0: Intel i82542 1000BASE-X Ethernet, rev. 3
wm0: interrupting at ioapic0 pin 18 (irq 10)
wm0: 64 word (6 address bits) MicroWire EEPROM
wm0: Ethernet address 00:08:c7:16:84:bb
wm0: 1000baseSX, 1000baseSX-FDX, auto

> Try netstat -i. You may also consider dropping the cisco's MTU to 1496...

   netstat -i reports no errors.  :-/

   Dropping the cisco's MTU on that interface to 1496?  While the
wm is still using 1500?  Or dropping the MTU of the vlans?

> It's more than likely a limitation of your particular wm(4) card.

   Right.  I see that.  Someone else made the comment about the 82543
supporting vlan, so mine doesn't.  :-(  I'm lucky it works to the
degree it does, I guess.  (*sigh*)  I'm not sure how to fix this.
Don't know how hard it would be to find another.

   Let me know if you have an inexpensive newer wm card with multimode
fiber interface.  :-)

   Thanks all!

                                  - Chris