All of lore.kernel.org
 help / color / mirror / Atom feed
* MTU question
@ 2012-04-20 16:15 Lloyd Standish
  2012-04-20 17:30 ` Rick Jones
  2012-04-20 17:41 ` Steven Kath
  0 siblings, 2 replies; 4+ messages in thread
From: Lloyd Standish @ 2012-04-20 16:15 UTC (permalink / raw)
  To: netfilter

Hi,

I have a question about MTU.  Suppose my PC's Ethernet interface is configured at 1500 MTU, and it connects to a router interface configured to use 1492 MTU.

Suppose my PC needs to send packets to the router.  Given these MTU settings, which of these 2 is correct?

(a) my PC will only send 1492 byte packets maximum (because the destination interface is only 1492), or
(b) the PC will send packets up to 1500 bytes in size, but the router will split the packets before sending them on

I have been assuming (a) is correct.

Thanks in advance.

--
Lloyd

^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: MTU question
  2012-04-20 16:15 MTU question Lloyd Standish
@ 2012-04-20 17:30 ` Rick Jones
  2012-04-22 14:32   ` Jan Engelhardt
  2012-04-20 17:41 ` Steven Kath
  1 sibling, 1 reply; 4+ messages in thread
From: Rick Jones @ 2012-04-20 17:30 UTC (permalink / raw)
  To: Lloyd Standish; +Cc: netfilter

On 04/20/2012 09:15 AM, Lloyd Standish wrote:
> Hi,
>
> I have a question about MTU. Suppose my PC's Ethernet interface is
> configured at 1500 MTU, and it connects to a router interface configured
> to use 1492 MTU.
>
> Suppose my PC needs to send packets to the router. Given these MTU
> settings, which of these 2 is correct?
>
> (a) my PC will only send 1492 byte packets maximum (because the
> destination interface is only 1492), or
> (b) the PC will send packets up to 1500 bytes in size, but the router
> will split the packets before sending them on
>
> I have been assuming (a) is correct.

I suspect the answer is some variation on:

c) Since the two systems are in the same broadcast domain, with no 
intervening router to perform PathMTU discovery, the PC will send IP 
datagrams up to 1500 bytes, which likely as not will be dropped by the 
"router" as being too large before it even gets to IP on the router.

I would expect that only if your "router" has a 1500 byte MTU on the 
interface facing the PC, and a 1492 byte MTU on the interface facing the 
"internet" would you see the PathMTU discovery activity which would 
cause the PC to adjust the effective MTU it was using to match 1492. 
Only if the IPv4 DF (Don't Fragment) bit is clear should the "router" 
split the packets.

rick jones

> Thanks in advance.
>
> --
> Lloyd
> --
> To unsubscribe from this list: send the line "unsubscribe netfilter" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at http://vger.kernel.org/majordomo-info.html


^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: MTU question
  2012-04-20 16:15 MTU question Lloyd Standish
  2012-04-20 17:30 ` Rick Jones
@ 2012-04-20 17:41 ` Steven Kath
  1 sibling, 0 replies; 4+ messages in thread
From: Steven Kath @ 2012-04-20 17:41 UTC (permalink / raw)
  To: Lloyd Standish; +Cc: netfilter

> I have a question about MTU.  Suppose my PC's Ethernet interface is
> configured at 1500 MTU, and it connects to a router interface
> configured to use 1492 MTU.
> 
> Suppose my PC needs to send packets to the router.  Given these MTU
> settings, which of these 2 is correct?
> 
> (a) my PC will only send 1492 byte packets maximum (because the
> destination interface is only 1492), or
> (b) the PC will send packets up to 1500 bytes in size, but the router
> will split the packets before sending them on
> 
> I have been assuming (a) is correct.

I'm assuming you mean 1492 is the MTU of the interface on the 
router facing the remote endpoint.  The MTU of the interface 
facing your PC is not directly relevant to your question.  

The answer depends on whether Path MTU discovery (RFC 1191) is 
enabled on the transmitting host.

Without PMTUD, sender has no information about what MTUs are 
in use on remote devices, and the behavior is usually (b). 

With PMTUD, the behavior is as described in RFC 1191. 

^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: MTU question
  2012-04-20 17:30 ` Rick Jones
@ 2012-04-22 14:32   ` Jan Engelhardt
  0 siblings, 0 replies; 4+ messages in thread
From: Jan Engelhardt @ 2012-04-22 14:32 UTC (permalink / raw)
  To: Rick Jones; +Cc: Lloyd Standish, netfilter


On Friday 2012-04-20 19:30, Rick Jones wrote:
>
> I suspect the answer is some variation on:
>
> c) Since the two systems are in the same broadcast domain, with no intervening
> router to perform PathMTU discovery, the PC will send IP datagrams up to 1500
> bytes, which likely as not will be dropped by the "router" as being too large
> before it even gets to IP on the router.
>
> I would expect that only if your "router" has a 1500 byte MTU on the interface
> facing the PC, and a 1492 byte MTU on the interface facing the "internet" would
> you see the PathMTU discovery activity which would cause the PC to adjust the
> effective MTU it was using to match 1492. Only if the IPv4 DF (Don't Fragment)
> bit is clear should the "router" split the packets.

Nope, if DF is set, no splitting is allowed and the appropriate
ICMP message is generated.

And exactly that is how PMTU is discovered.

^ permalink raw reply	[flat|nested] 4+ messages in thread

end of thread, other threads:[~2012-04-22 14:32 UTC | newest]

Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2012-04-20 16:15 MTU question Lloyd Standish
2012-04-20 17:30 ` Rick Jones
2012-04-22 14:32   ` Jan Engelhardt
2012-04-20 17:41 ` Steven Kath

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.