Re: [Xen-users] Changing MTU of vif

[Tom Brown <xensource.com@xxxxxxxxxxxxxxxxxxx>]

> AoE with Xen works perfectly for me. The only problem is that it is rather 
> challenging to set up. You should be a pretty experienced sysadmin to get it 
> going.
> I have set up a website at http://xenaoe.org where I am documenting my whole 
> setup and promoting the idea. I still need to get my initrd up there (pretty
I'd like to have heard your thoughts on the various AOE target 
implimentations (vblade, kernel-vblade, qaoed, etc...)
----

I did come up with a crude fix for your vif-bridge problem...

   log debug "Successful vif-bridge $command for $vif, bridge $bridge."
   if [ "$command" == "online" ]
   then
     success
   fi

after the "success" command, I added the following

   (sleep 10; ifconfig "$vif" mtu 5200) &

e.g.

   log debug "Successful vif-bridge $command for $vif, bridge $bridge."
   if [ "$command" == "online" ]
   then
     success
     do_without_error ifconfig "$vif" mtu 5200
     (sleep 10; ifconfig "$vif" mtu 5200) &
   fi

If you're not familiar with the syntax, the parenthesis cause a
another shell to be forked with the enclosed commands, and the
ampersand puts it in the background, so the main vif-bridge
script basically forks a child to sleep and tweak the mtu after a
delay.

It is not pretty, it is not clean, but then having a script like
the following run every 60 seconds isn't clean or pretty either:

#!/bin/bash
#
cd /sys/class/net/xenbr0/brif/ 
for f in * ; do
   echo ifconfig $f mtu 5200;
   ifconfig $f mtu 5200; 
done
(why did I pick 5200? Because I wanted the system to use 4096
byte aoe transfers. )

(Oh, and from what I have seen, the MTU on the bridge device doesn't need 
to be set... it is automatically the smallest MTU of any of the interfaces 
attached to the bridge.)
-Tom


_______________________________________________
Xen-users mailing list
Xen-users@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-users