Sidebar to Apple OS X SMB issues across VPN

David Collier-Brown David.Collier-Brown at Sun.COM
Tue Oct 25 12:00:52 GMT 2005

  There are often two settings that give good speed: just
small enough to fit into an MTU, and really, really big.

  The former gets efficiencies if fragmentation and
errors cause retransmissions, but it assumes the application
can process many small reads/writes quickly.

  The latter gets efficiencies if the application
works best with few large writes, but assumes the
IP stack will do the packetization elegantly.

  You're using the latter, which often works well
with samba. I'd try unsetting the "Ignore DF (Don't Fragment)
Bit" at the offsite end and see if it has an effect. 
Mismatching them is probably not good.


Dan Tappin wrote:
> My brain hurts after reading this :)...
> Being that this my system being dissected I will add some comments.
> I played with the:
>  - Xserve's MTU (currently 1492 - default 1500)
>  - the Xserve's Sonicwall's MTU (1500)
> Here is the options from the Sonicwall's ethernet config page:
>  - (checked) Fragment non-VPN outbound packets larger than WAN MTU
>  - (not checked) Ignore DF (Don't Fragment) Bit
>  - WAN MTU: 1500
> hmmm... I just double checked the offsite's Sonicwall:
>  - (checked) Fragment non-VPN outbound packets larger than WAN MTU
>  - (checked) Ignore DF (Don't Fragment) Bit
>  - WAN MTU: 1500
> So if one device is passing unfragmented data and the other is  
> fragmenting it any way (remember I know nothing about TCP/IP -  forgive 
> my ignorance) could this be the issue?
> I have made both the same (not checked).  I will double check the  speed 
> issue and then also try both checked.  Before I do any  suggestions as 
> to which one would be better?  Same goes for a MTU  setting?
> Dan T
> On Oct 24, 2005, at 2:56 PM, Christopher R. Hertel wrote:
>> Dave,
>> The continuation messages occur whenever the SMB message is larger  than
>> the MTU on the link.  It's the TCP/IP stack that breaks the large  SMB 
>> down
>> into smaller chunks for transmission.
> <<snip>>

David Collier-Brown,      | Always do right. This will gratify
Sun Microsystems, Toronto | some people and astonish the rest
davecb at     |                      -- Mark Twain
(416) 263-5733 (x65733)   |

More information about the samba-technical mailing list