> Others, I used 1500/1500, it all depends on the ISP. Smaller, isnt
It has nothing to do with the ISP, my ISP supports all of them just fine.
The advantage to using the small MTU/MRU is better interactive response
when the pipe is full to overflowing. I have already reported that it
works fine if I change the PPP MTU/MRU to 1500. My question is:
Given:
+------------------+
+--------+ | Linux w/ masq | +-----------+
| | | | | |
| World <--PPP--> Real IP <--eth0--> Whatever |
| | | mtu=296 | | |
+--------+ | | +-----------+
+------------------+
Problem: 'Whatever' negotiates <mss=1460> with 'World'. Path-MTU
discovery is broken somewhere in 'World'.
Proposed workaround: The masq machine *knows* it is masquerading over an
<mss=256> link, and is already messing about in the packet headers, would
it *work* for the masq machine to *change* the <mss=1460> to <mss=256>,
and if not, why not? I don't want to start hacking in the code if there
are fundamental problems. For instance, *does* the masq code in fact have
access to the MTU of the PPP link in the above diagram? Would reducing
the <mss=1460> when forwarding break anything else? Are there any other
drawbacks to reducing the mss from 1460 to 256, given that the forwarding
machine *knows* (I hope) that it is going over the smaller PPP?
-Tom
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
For daily digest info, email [EMAIL PROTECTED]
