[c-nsp] Scheduling daily reload

Tom Storey tom at snnap.net
Tue Jan 1 21:27:31 EST 2008


On 02/01/2008, at 12:25 PM, Aaron R wrote:

> Hi Roland,
>
> The problem is the ISP tends to make changes at the various  
> exchanges (The
> links are ADSL) and this sometimes causes a problem whereby no  
> packets will
> transmit / receive however the router still believes that the link  
> is up.
> Maybe I need to look into setting some kind of keep alive feature  
> over the
> ATM interface perhaps.
>
> Cheers,
>
> Aaron.
>

I couldnt imagine that an ISP would make daily changes to the effect  
that it would cause your services to stop working and require a  
reload. Maybe I expect too high operational standards? ;-)

Perhaps there is some other issue worth investigating.

Are the services layer 3 bridged by any chance, i.e. not PPP based?

Perhaps it is an issue with ARP at the ISP end. If you dont send any  
traffic for a period of time and your ARP entry times out, this would  
cause data flow on a bridged service to stop, however, the ARP entry  
should return at their end once you start generating a bit of traffic,  
so that should be easily fixed.

In the case above, something like an SLA monitor which simply pings an  
IP address accross the link, say, once a minute, should ensure that  
ARP stays "fresh" at the ISP end. It is always possible that the  
firmware their DSLAMs or hardware at their PoP runs has a bug and  
doesnt properly re-establish an ARP entry for your service(s).

Otherwise, try a different IOS version (newer IOS images contain ADSL  
controller firmware updates too, just incase its a local bug), and  
failing all of that, log a fault with the provider and see if you can  
get them to fix their end. You shouldnt be needing to implement  
anything at your end to relieve yourself from their issues. But then  
again, am I expecting too high standards? :-)


More information about the cisco-nsp mailing list