Hey everyone, I am trying to do some NIC Bondig but do not have kernel-doc installed - I know there is a bonding.txt file in there if it is installed. My question is how to I install kernel-doc? I am running Redhat EL4 2.6.9-42.0.3.ELsmp

Thanks in advance

why do you even need it? just check the howto online.

http://www.google.co.uk/search?q=bon...en-GB:official

Hey thanks for the link but it is a bit too vague for me since I have never done it before. I do not know if I have support in the kernel already etc. I need kind of a noob howto. Any ideas?
I installed the kernel-doc and I a mreading through that..but it is losing me here and there. I just want to bond two 1 gb nic's for 2gb, they are both going to the same switch. This is a RHEL 4 box...

Thanks

you do have support already under redhat, it's always there. just read through one of those docs and you'll see that it IS noob level. it's just a few tiny commands you need to use, you can do it all just by adding a relevant config under /etc/sysconfig/network-scripts/ as detailed very clearly in the first hit on google.

Hey, thanks for your help on this! But I am still a little confused. My system does not have a modules.conf or conf.modules file in /etc/ and I searched for them and there are no files o nthe system with these names. This is step three in the howto. Can I skip this step in EL4?

Thanks.

it's modprobe.conf nowadays, try this: http://www.cyberciti.biz/tips/linux-...interface.html

it's great being a human search engine!

Hey, thanks for your help and advice. I got it all working - I think. I cannot use ethtool on bond0 so how do I knwo if it is now a 2gb connection? I can still run ethtool on the eth0 and eht1 interfaces and they show 1000 Full. So how do I test taht I actually have a bond at 2 GB?

Thanks again!

check /proc/net/bonding/bond0

Ok cool! Check thisout:
more bond0
Ethernet Channel Bonding Driver: v2.6.3 (June 8, 2005)

Bonding Mode: load balancing (round-robin)
MII Status: up
MII Polling Interval (ms): 0
Up Delay (ms): 0
Down Delay (ms): 0

Slave Interface: eth0
MII Status: up
Link Failure Count: 0
Permanent HW addr: 00:14:c2:c3:bf:cb

Slave Interface: eth1
MII Status: up
Link Failure Count: 0
Permanent HW addr: 00:14:c2:c3:bf:ca

It looks like I have it configured for load balancing? Does that translate into 1 2GB pipe?

Thanks again, you are being a great help!!

well not necessarily. the size of the links doesn't really affect the bonded interface. there *may* be a requirement that they are the same size links, but i'm not sure. what you have there though is a RR based load balance link, which will alternate between links at an individual packet level. so you should see a general spread of load between the nics for outbound traffic. the inbound traffic though, i'd assume that it will only come down the master nic of the two, presumably eth0, as it can only respond to an arp request on a segment with one mac address per entry router. so if traffic comes from behind a single router, one nic would receive all that traffic, although i wouldn't be suprised if which nic address was returned during an arp request did change periodically, so you'd see a swing back and forth.

how bout this?

Issue:
How do I configure different modes on bonded interfaces (bonding) in Red Hat Enterprise Linux version 3?
Resolution:
Red Hat Enterprise Linux version 3 bonding supports 7 possible "modes" (0-6) for bonded interfaces. These modes determine the way in which traffic sent out of the bonded interface is dispersed over the real interfaces.

To configure the mode of a bonding interface the mode option for the kernel module supporting that interface in /etc/modules.conf must be set. The example lines in /etc/modules.conf could look like the following in order to set the mode of the bond0 interface to 0 and the mode of the bond1 interface to 1.


alias bond0 bonding
options bond0 mode=0 miimon=100
alias bond1 bonding
options bond1 -o bonding1 mode=1 miimon=100


These options are put into place when the module is loaded into the kernel. So in the above example if the mode is changed and the /etc/sysconfig/network-scripts/ are correct the following lines will reload the kernel module with the new mode settings.

service network stop
rmmod bonding
rmmod bonding1
service network start


Note: On version 4 its modprobe.conf not modules.conf

http://kbase.redhat.com/faq/FAQ_79_3758.shtm
for problems with red hat enterprise linux you can always check the knowledgebase
http://kbase.redhat.com/faq/

Check this out:
here is my modprobe.conf
#Bonding Config
alias bond0 bonding
options bond0 mode=0 miimon=100

and here is cat /proc/net/bonding/bond0
Ethernet Channel Bonding Driver: v2.6.3 (June 8, 2005)

Bonding Mode: load balancing (round-robin)
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: 1
Permanent HW addr: 00:14:c2:c3:bf:cb

Slave Interface: eth1
MII Status: up
Link Failure Count: 0
Permanent HW addr: 00:14:c2:c3:bf:ca
[zmeekins@bl-mediaagent ~]$

is this more of a teamed config?

Thanks

well that's just exactly the same. not really sure what the other guy was getting at with their post, but outbound traffic should have a 2gb bandwidth roughly speaking. if you want to look ing into more advanced LB methods, e.g. 802.3ad (mode 4) check the linux bonding howto or a good description of ll LB modes

Cool man thanks, I will look into it. This box is a backup server so I really need to have the inbound able to handle 2 gb in bandwidth.

Thanks for all the help!

well as a final note, I don't think any generic LB method will be able to do that for you. to make use of inbound capacity, you really need to have just an intelligent switch as the server, something like a cisco switch can do the proper 802.3ad load balancing, an send the data down to the box across multiple nics.