Thu Feb 13 11:20:25 NZDT 2003
>How I've thought this should work:
>- set a route for both modems, giving the Tempest modem a higher metric
>(this might not be necessary?)
A higher metric will cause it to be used less preferentially. As you
don't have load balancing enabled, this wll mean its not used at all -
until the route with the lower metric is disabled.
Set the route to your backup connection as having a higher metric
>- periodically discover the connection state on the Tempest connection using
>a script of some sort
> - if fine, do nothing
> - if not fine, set default route to Jetstream modem and start script
>watching Tempest connection
The above is fine, except all you need to do is delete the default route
- it'll then start using the route to the backup link
>The script watching the Tempest connection needs to see if the connection is
>up, and when connectivity returns, change the route back to the original
One cavaet is that the IP you are checking for the tempest connection
*must* have a route out the tempest link. Otherwise when you drop the
default gw, you find you can ping the ip again (through the backup) and
you bring it back up again... route flap!
Eg, add a route to www.clear.net.nz via the tempest gw, and ping that.
It'll keep pinging out over the tempest link. If the link goes down,
it'll still try to ping out over the tempest link, and so when it starts
working again your link is fine
>Would a ping test fired by something like Nagios be enough for this task?
>The modems support RIP, but I'm not sure what I'd need to do with the Linux
>machine to use it, and if it's overkill? Am I missing anything?
Yes, Nagios would be fine - although I had some issues getting nagios'
eventhandlers to do this nicely and wrote a cron scrip to do the same
thing. It even waits a few iterations before declaring the route dead.
I'll add a wiki entry now
More information about the wlug
NOTICE: This is an archive of a public mailing list. The University of Waikato is not responsible for its contents.