This presentation, given at the SHARE Orlando meeting last February, might be
of interest:
"Networking with Linux® on System z", session 9267, given by Steffen Thoss
([email protected]). He discusses network bonding of multiple NICs in some detail, and
gives example configurations for SLES 10.
If you can't find it on the SHARE site, let me know and I'll send you a copy
offline.
DJ
Harder, Pieter wrote:
Here are my results:
TSMSERV:~ # cat /proc/net/bonding/bond0
Ethernet Channel Bonding Driver: v3.0.3 (March 23, 2006)
Bonding Mode: load balancing (xor)
Transmit Hash Policy: layer2 (0)
MII Status: up
MII Polling Interval (ms): 100
Up Delay (ms): 0
Down Delay (ms): 0
Slave Interface: eth0
MII Status: up
Link Failure Count: 0
Permanent HW addr: 02:00:00:02:03:96
Slave Interface: eth1
MII Status: up
Link Failure Count: 0
Permanent HW addr: 02:00:00:02:03:96
TSMSERV:~ # ifconfig
bond0 Link encap:Ethernet HWaddr 02:00:00:02:03:96
inet addr:10.2.3.150 Bcast:10.2.3.255 Mask:255.255.255.0
inet6 addr: fe80::ff:fe02:396/64 Scope:Link
UP BROADCAST RUNNING MASTER MULTICAST MTU:8992 Metric:1
RX packets:1731908001 errors:0 dropped:0 overruns:0 frame:0
TX packets:614119188 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:0
RX bytes:2543338706705 (2425516.8 Mb) TX bytes:49090109816 (46815.9
Mb)
eth0 Link encap:Ethernet HWaddr 02:00:00:02:03:96
UP BROADCAST RUNNING SLAVE MULTICAST MTU:8992 Metric:1
RX packets:1731753036 errors:0 dropped:0 overruns:0 frame:0
TX packets:613696957 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:2543333960229 (2425512.2 Mb) TX bytes:48870854378 (46606.8
Mb)
eth1 Link encap:Ethernet HWaddr 02:00:00:02:03:96
UP BROADCAST RUNNING SLAVE MULTICAST MTU:8992 Metric:1
RX packets:154965 errors:0 dropped:0 overruns:0 frame:0
TX packets:422231 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:4746476 (4.5 Mb) TX bytes:219255438 (209.0 Mb)
lo Link encap:Local Loopback
inet addr:127.0.0.1 Mask:255.0.0.0
inet6 addr: ::1/128 Scope:Host
UP LOOPBACK RUNNING MTU:16436 Metric:1
RX packets:9791 errors:0 dropped:0 overruns:0 frame:0
TX packets:9791 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:0
RX bytes:58948335 (56.2 Mb) TX bytes:58948335 (56.2 Mb)
TSMSERV:~ # vmcp q nic details
Adapter 0C00.P00 Type: QDIO Name: 0 Devices: 3
MAC: 02-00-00-02-03-96 VSWITCH: SYSTEM QDIOL
RX Packets: 1731753469 Discarded: 0 Errors: 0
TX Packets: 613696999 Discarded: 0 Errors: 0
RX Bytes: 2567578537583 TX Bytes: 48870861170
Connection Name: HALLOLE State: Session Established
Device: 0C00 Unit: 000 Role: CTL-READ
Device: 0C01 Unit: 001 Role: CTL-WRITE
Device: 0C02 Unit: 002 Role: DATA vPort: 0079 Index: 0079
VLAN: 0203
Options: Ethernet Broadcast
Unicast MAC Addresses:
02-00-00-02-03-96
Multicast MAC Addresses:
01-00-5E-00-00-01
01-00-5E-00-01-16
01-00-5E-7F-FF-FD
33-33-00-00-00-01
33-33-FF-02-03-96
Adapter 1C00.P00 Type: QDIO Name: 0 Devices: 3
MAC: 02-00-00-00-00-06 VSWITCH: SYSTEM QDIOL
RX Packets: 155365 Discarded: 0 Errors: 0
TX Packets: 422287 Discarded: 0 Errors: 0
RX Bytes: 6974051 TX Bytes: 219341818
Connection Name: HALLOLE State: Session Established
Device: 1C00 Unit: 000 Role: CTL-READ
Device: 1C01 Unit: 001 Role: CTL-WRITE
Device: 1C02 Unit: 002 Role: DATA vPort: 0080 Index: 0080
VLAN: 0203
Options: Ethernet Broadcast
Unicast MAC Addresses:
02-00-00-02-03-96
Multicast MAC Addresses:
01-00-5E-00-00-01
01-00-5E-00-01-16
01-00-5E-7F-FF-FD
33-33-00-00-00-01
33-33-FF-02-03-96
As you can see neither the incoming nor the outgoing data are balanced. We all
expected that for the incoming data, as the switch does have no information
about the bonding. For the outgoing data it is probably caused by the fact that
Transmit Hash Policy: layer2 (0) to one connecting switch results in mostly the
same hash. It would be interesting to see Hash Policy layer2+3, but that is not
available on the current level of the bonding driver in SLES10. On the
performance side of things (as displayed by PTK and ESALPS) the addition of the
bonding drivers doesn't seem to have caused any noticable effects. Next week I
hope to try the same thing with a pair of real OSA's and a switch that does
know about bonding and see what happens then. And a refresh of the bonding
driver would come in handy as well (Mark, can you put that on someone's list?)
Best regards,
Pieter Harder
[email protected]
tel +31-73-6837133 / +31-6-47272537
Brabant Water N.V.
Postbus 1068
5200 BC 's-Hertogenbosch
http://www.brabantwater.nl
Handelsregister: 16005077
----------------------------------------------------------------------
For LINUX-390 subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO LINUX-390 or visit
http://www.marist.edu/htbin/wlvindex?LINUX-390
--
DJ
V/Soft
z/VM and mainframe Linux expertise, training,
consulting, and software development
www.vsoft-software.com
----------------------------------------------------------------------
For LINUX-390 subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO LINUX-390 or visit
http://www.marist.edu/htbin/wlvindex?LINUX-390
