IPTABLES and load balancing multiple ISPs

pablouruguay, this tutorial is for routed, not NATed links. For NATed it doesn't fit well. This problem described here: http://linux-ip.net/html/adv-multi-internet.html#adv-multi-internet-inbound (wesbird posted it in his question :)


wesbird, I suppose you are using kernel 2.6.15?
Read here how to install 'random' module. Probably you have old version of patch-o-magic. Read here for more details: http://www.linuxjournal.com/node/7180/
http://www.netfilter.org/patch-o-matic/pom-base.html#pom-base-random

Running POM is simple; download the latest Patch-o-matic tarball from the directory /pub/patch-o-matic on ftp.netfilter.org, restore it on your system and run the following command while logged in as root. Make sure to give the correct kernel source directory name as the value of the KERNEL_DIR parameter:

KERNEL_DIR=/usr/src/linux-2.6.15 ./runme random

Hi again,

Am I making some silly mistake here?

did a make clean on the kernel,
tried ./runme random on the correct kernel source tree

it said that the patch was already applied
 
rebuilt the kernel and I'm still getting this:

Try `iptables -h' or 'iptables --help' for more information.
iptables v1.3.5: Couldn't load match `random':/usr/lib/iptables/libipt_random.so: cannot open shared object file: No such file or directory

Hi again,

ftp.netfilter.org/pub/patch-o-matic-ng/snapshot/patch-o-matic-ng-20060427.tar.bz2 has allowed me select it, but I've still got a couple of problems.  

iptables: Unknown error 4294967295

>> After patching kernel, don't forget to run 'make menuconfig' a to toggle 'random match' and 'nth match' modules in appropriate branches.

I've just spotted that I've still got a problem with the build.  I'm using make gconfig as opposed to menuconfig - it didn't seem to do the appropriate update.    I just found nth and random under Networking->Networking Support->Networking Options->Network Packet Filtering and they weren't checked.

Will post again tomorrow after attempting another rebuild.

Thanks,

We've also applied this patch for dead gateway detection which seems to improve things somewhat: http://www.ssi.bg/~ja/

The dead gateway detection is really cool and now the automatic failover is working too - it still takes a minute or so, but that's quicker than I can fix it at 3am - not just from disconencting eth1 or eth2, but also line failure of the DSL boxes.

Thanks for your help.

These are the values changed:

root@darkstar:/proc/sys/net/ipv4/route# ls -l
total 0
-rw-r--r--  1 root root 0 2006-05-25 16:21 error_burst
-rw-r--r--  1 root root 0 2006-05-25 16:21 error_cost
--w-------  1 root root 0 2006-05-25 16:21 flush
-rw-r--r--  1 root root 0 2006-05-25 16:21 gc_elasticity
-rw-r--r--  1 root root 0 2006-05-25 16:21 gc_interval
-rw-r--r--  1 root root 0 2006-05-25 16:21 gc_min_interval
-rw-r--r--  1 root root 0 2006-05-25 16:21 gc_min_interval_ms
-rw-r--r--  1 root root 0 2006-05-25 16:21 gc_thresh
-rw-r--r--  1 root root 0 2006-05-25 16:21 gc_timeout
-rw-r--r--  1 root root 0 2006-05-25 16:21 max_delay
-rw-r--r--  1 root root 0 2006-05-25 16:21 max_size
-rw-r--r--  1 root root 0 2006-05-25 16:21 min_adv_mss
-rw-r--r--  1 root root 0 2006-05-25 16:21 min_delay
-rw-r--r--  1 root root 0 2006-05-25 16:21 min_pmtu
-rw-r--r--  1 root root 0 2006-05-25 16:21 mtu_expires
-rw-r--r--  1 root root 0 2006-05-25 16:21 redirect_load
-rw-r--r--  1 root root 0 2006-05-25 16:21 redirect_number
-rw-r--r--  1 root root 0 2006-05-25 16:21 redirect_silence
-rw-r--r--  1 root root 0 2006-05-25 16:21 secret_interval
root@pcgate:/proc/sys/net/ipv4/route# cat *
1250
250
cat: flush: Operation not permitted
5
60
0
0
2048
0
10
32768
256
2
552
600
5
9
5120
600
root@darkstar:/proc/sys/net/ipv4/route#