I was wondering if anyone out there has had success running lustre atop
a ''vlan-tagged'' interface with lnet? If so please let me know
what you did!
Here''s the description of the problem..
I''m trying to bind lustre to the 112 vlan which hangs off eth0.
The lnet module is being loaded with `options lnet
networks="tcp(eth0.112)"`
with no error but the mds won''t communicate with the ost. It gives the
following
error:
Jul 14 17:29:50 mds00p kernel: LustreError:
3266:0:(socklnd_cb.c:801:ksocknal_launch_packet()) No usable routes to
12345-10.10.100.110@tcp
So I have a feeling that the mds lnet/socknald is attaching directly to
eth0
and not the vlan interface, eth0.112. Since eth0 has no ip address the
''no routes''
msg seems to make sense. Also when performing a tcpdump on eth0 it is
apparent
that the mds is attempting to communicate via the eth0 interface instead
of eth0.112.
paul
(root@mds00p:~)# ifconfig
eth0 Link encap:Ethernet HWaddr 00:04:23:BD:68:34
UP BROADCAST RUNNING MULTICAST MTU:9000 Metric:1
RX packets:24886 errors:0 dropped:0 overruns:0 frame:0
TX packets:18993 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:1911216 (1.8 MiB) TX bytes:5161896 (4.9 MiB)
Base address:0x3000 Memory:d0140000-d0160000
eth0.90 Link encap:Ethernet HWaddr 00:04:23:BD:68:34
inet addr:10.10.101.108 Bcast:10.10.101.255 Mask:255.255.255.0
UP BROADCAST RUNNING MULTICAST MTU:9000 Metric:1
RX packets:63 errors:0 dropped:0 overruns:0 frame:0
TX packets:4 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:0
RX bytes:3150 (3.0 KiB) TX bytes:168 (168.0 b)
eth0.112 Link encap:Ethernet HWaddr 00:04:23:BD:68:34
inet addr:10.10.100.108 Bcast:10.10.100.255 Mask:255.255.255.0
UP BROADCAST RUNNING MULTICAST MTU:9000 Metric:1
RX packets:24662 errors:0 dropped:0 overruns:0 frame:0
TX packets:18996 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:0
RX bytes:1353918 (1.2 MiB) TX bytes:5008656 (4.7 MiB)
# tcpdump -i eth0
17:32:45.735653 arp who-has oss00p tell mds00p
17:32:45.735874 IP mds00p.1023 > oss00p.988: S 2086306321:2086306321(0)
win 17920 <mss 8960,sackOK,timestamp 4548839 0,nop,wscale 9>
17:32:45.736133 IP mds00p.1023 > oss00p.988: . ack 2501376205 win 35
<nop,nop,timestamp 4548840 11529996>
17:32:45.736145 IP mds00p.1023 > oss00p.988: P 0:16(16) ack 1 win 35
<nop,nop,timestamp 4548840 11529996>
17:32:45.736361 IP mds00p.1023 > oss00p.988: F 16:16(0) ack 1 win 35
<nop,nop,timestamp 4548840 11529996>
17:32:45.736755 IP mds00p.1023 > oss00p.988: . ack 2 win 34
<nop,nop,timestamp 4548841 11529997>