[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

strange IP question


I'm running into a continuous problem with a box that has several IP
numbers on an Ethernet interface. The problem persisted across the last
three releases of OpenBSD at least (currently: 3.6-stable), and is as

# /sbin/ifconfig fxp3
        address: 00:30:48:51:58:55
        media: Ethernet autoselect (100baseTX full-duplex)
        status: active
        inet netmask 0xffff0000 broadcast
        inet6 fe80::230:48ff:fe51:5855%fxp3 prefixlen 64 scopeid 0x4
        inet netmask 0xffff0000 broadcast
        inet netmask 0xfffffff8 broadcast

The 180.180 can be safely ignored - it's an artefact that's bound to go
away soon.

The packet filter allows 172.something and 180.180 to communicate, and
the 10.something to communicate with other 10.somthings.

Now, on pflog0, I see this:

11:25:35.937147 > icmp: unreachable - need to frag (mtu 1378)
11:25:41.536690 > icmp: unreachable - need to frag (mtu 1378)

Why does the (OpenBSD) router use instead of for
it's sender address, and how can i force it to use the when
talking to the 172.16. network?

So far I assumed that it would talk to the 10.x peers using it's 10.x
address, and to 172.x peers using it's 172.x address, and so on.