[Rack] WAN loadbalancing rule updates
Ben Kochie
ben at nerp.net
Mon Sep 3 01:36:07 UTC 2012
We noticed a lot of instability this afternoon. From what I could tell it
was all related to monkeybrains packet loss.
I tested the configuration by unplugging monkeybrains and it failed over
to sonic.
The next thing on my todo list is to get bikeshed to syslog to minotaur
properly so we know about failover events.
I also think I figured out some of the issues with doing port forwarding
on bikeshed. I'm going to poke around with the policy routing tables.
-ben
On Sun, 2 Sep 2012, Ben Kochie wrote:
> So I figured out some stuff with the vyatta config to help with WAN load
> balancing.
>
> So here are how the interfaces are configured:
>
> interface-health eth0 {
> failure-count 5
> nexthop 199.188.193.1
> success-count 1
> test 10 {
> resp-time 5
> target 208.90.215.1
> ttl-limit 1
> type ping
> }
> }
> interface-health eth2 {
> failure-count 4
> nexthop 75.101.62.1
> success-count 1
> test 10 {
> resp-time 5
> target 70.36.228.61
> ttl-limit 1
> type ping
> }
>
> They have a 5 second ICMP ping test to their local gateways.
>
> I updated the rules a bit.
>
> rule 10 {
> destination {
> port 22
> }
> inbound-interface eth1
> interface eth2 {
> }
> protocol tcp
> }
> rule 11 {
> destination {
> port 22
> }
> inbound-interface eth1
> interface eth0 {
> }
> protocol tcp
> }
>
> These two rules mean that SSH is configured to stick to eth2 (sonic) unless
> it's down.
>
> rule 20 {
> failover
> inbound-interface eth1
> interface eth0 {
> weight 100
> }
> interface eth2 {
> weight 1
> }
> protocol all
> }
>
> This sets the rest of the traffic to failover with a weight of 100 eth0
> (monkeybrains)
>
> -ben
> _______________________________________________
> Rack mailing list
> Rack at lists.noisebridge.net
> https://www.noisebridge.net/mailman/listinfo/rack
>
More information about the Rack
mailing list