[Bridge] Re: aoe/vblade on "localhost"

devzero at web.de devzero at web.de
Sun Mar 12 05:10:07 PST 2006


some more info:

> afaik, after starting, the bridge is in some "learning" state, to see what is on the "left side" and on the "right side".
> i`m somewhat unsure if AoE/vblade are sending out enough (or the right?) packets to make the bridge let them go through (aoe and vblade get aware of each other via broadcast) 
> 
> can somebody give an advice on what needs to be taken into account here and what could be done to debug this?
> regards

actually, the broadcasts of AoE driver get through from dummy1 to dummy0 - i see 2 packets on dummy0 and those 2 being forwarded on dummy1

vserver2:~ # tcpdump -i dummy0 &
[1] 11302
vserver2:~ # tcpdump: verbose output suppressed, use -v or -vv for full protocol decode
listening on dummy0, link-type EN10MB (Ethernet), capture size 96 bytes

vserver2:~ # tcpdump -i dummy1 &
[2] 11303
vserver2:~ # tcpdump: verbose output suppressed, use -v or -vv for full protocol decode
listening on dummy1, link-type EN10MB (Ethernet), capture size 96 bytes

14:02:09.887410 6e:d9:cf:4c:9d:df > Broadcast, ethertype Unknown (0x88a2), length 60:
        0x0000:  1000 ffff ff01 0000 0000 0000 0000 0000  ................
        0x0010:  0000 0000 0000 0000 0000 0000 0000 0000  ................
        0x0020:  0000 0000 0000 0000 0000 0000 0000       ..............
14:02:09.888004 ca:90:06:04:f6:ec > Broadcast, ethertype Unknown (0x88a2), length 60:
        0x0000:  1000 ffff ff01 0000 0000 0000 0000 0000  ................
        0x0010:  0000 0000 0000 0000 0000 0000 0000 0000  ................
        0x0020:  0000 0000 0000 0000 0000 0000 0000       ..............
14:02:09.886699 6e:d9:cf:4c:9d:df > Broadcast, ethertype Unknown (0x88a2), length 60:
        0x0000:  1000 ffff ff01 0000 0000 0000 0000 0000  ................
        0x0010:  0000 0000 0000 0000 0000 0000 0000 0000  ................
        0x0020:  0000 0000 0000 0000 0000 0000 0000       ..............
14:02:09.888278 6e:d9:cf:4c:9d:df > Broadcast, ethertype Unknown (0x88a2), length 60:
        0x0000:  1000 ffff ff01 0000 0000 0000 0000 0000  ................
        0x0010:  0000 0000 0000 0000 0000 0000 0000 0000  ................
        0x0020:  0000 0000 0000 0000 0000 0000 0000       ..............


but - the broadcasts from vblade (connected to dummy0) can only be seen on dummy0, but not on dummy1:


vserver2:~ # tcpdump -i dummy0
tcpdump: verbose output suppressed, use -v or -vv for full protocol decode
listening on dummy0, link-type EN10MB (Ethernet), capture size 96 bytes

14:04:33.825077 ca:90:06:04:f6:ec > Broadcast, ethertype Unknown (0x88a2), length 32:
        0x0000:  0800 0000 0001 0000 0000 0020 400a 0010  ............ at ...
        0x0010:  0000                                     ..

it seems the bridge isn`t forwarding them from dummy0 to dummy1

i have setup the bridge 2 or 3 times - but it doesn`t seem to work as expected.

i`m on  SuSE 9.2 with kernel 2.6.8-24.19-default 

regards
roland



> -----Ursprüngliche Nachricht-----
> Von: devzero at web.de
> Gesendet: 12.03.06 12:38:49
> An: bridge at osdl.org
> Betreff: aoe/vblade on "localhost"


> hello !
> 
> i try to use a network technology on one single host, which wasn`t designed for that. 
> 
> to give a short overview of what i`m talking about:
> 
> AoE is just like a "networked blockdevice" (just like nbd/enbd) - but without tcp/ip. 
> AoE kernel driver is the "client end" (see this like an iSCSI initiator) - and an etherblade storage appliance is the "server end" (see this like an iSCSI target)
> there is a software equivalent to etherblade storage which is called vblade (part of the aoetools project) - or vblade-kernel (another implementation, but done as a kernel module)
> 
> a very nice article about what AoE is can be found at http://www.linuxjournal.com/comment/reply/8149
> 
> what i want to do is nicely described in this bugreport by some other person:
> 
> http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=336772
> 
> i want to connect a AoE driver to vblade on _one_ host and i have tried that mentioned "clever juggling of network devices" on my system, but i didn`t succedd.
> 
> some coraid employee recommended trying linux-bridging - so i created two dummy interfaces (module dummy.ko), created a bridge "aoe-bridge" and added dummy0 and dummy1 to that bridge.
> 
> then i "connected" AoE-driver to dummy0 and vblade server process to dummy1 and would have expected that this should work.
> 
> unfortunately, it failed to work and from what it seems it wasn`t a problem with AoE or vblade but with the bridge, because i couldn`t see (tcpdump) packets going through into both directions. 
> 
> afaik, after starting, the bridge is in some "learning" state, to see what is on the "left side" and on the "right side".
> i`m somewhat unsure if AoE/vblade are sending out enough (or the right?) packets to make the bridge let them go through (aoe and vblade get aware of each other via broadcast) 
> 
> can somebody give an advice on what needs to be taken into account here and what could be done to debug this?
> regards
> 
> roland k. 
> systems engineer


______________________________________________________________
Verschicken Sie romantische, coole und witzige Bilder per SMS!
Jetzt bei WEB.DE FreeMail: http://f.web.de/?mc=021193





More information about the Bridge mailing list