LVS/TUN MTU problem, missing hints in documentation

[jarol1@seznam.cz]

Joseph Mack NA3T wrote:
> On Thu, 7 Dec 2006, jarol1 at seznam.cz wrote:
>
>> Today I ran into problem of MTU in LVS/Tunneling.
>>
>> Small packets were forwarded to real servers without problems, but 
>> the bigger ones weren't and TCP retransmissions occurred. I noticed 
>> the problem dissapeared when I switched to LVS/DR so this gave me 
>> hint to where the problem might be. MTU 1480 had to be set on the 
>> outgoing interface of realservers with tunl0 having standard 1500. 
>> Directors have 1500 on all interfaces. This way TCP syn ack contained 
>> correct MTU and client didn't send big packets that were discarded on 
>> director anymore. IP header is 20 bytes long so 1480 is the maximum 
>> value that works.
>>
>> It isn't mentioned in documentation that this problem will occur if 
>> tunneling is used
>
> http://www.austintek.com/LVS/LVS-HOWTO/HOWTO/LVS-HOWTO.LVS-Tun.html#MTU
>
> This was a hard problem to figure out and solve. People worked on this 
> for about 2yrs before finding a satisfactory solution. The problem is 
> that the ipip code in Linux doesn't handle fragmentation (or handle it 
> properly).
>
> You did well to come up with a solution in one go.
>
> Joe
>
I was thinking a sort of warning could be added either to 
http://www.linuxvirtualserver.org/VS-IPTunneling.html or to the 
ultramonkey website. There could be a link to the 
http://www.austintek.com/LVS/LVS-HOWTO/HOWTO/LVS-HOWTO.LVS-Tun.html#MTU 
on how to deal with it.

Jaro