[c-nsp] Dialup problems on a AS5300

[Justin Shore]

Vinny Abello wrote:
> Hi Justin,
> 
> I'm assuming you have MICA modems, and you didn't mention which IOS version you are running. It could actually be bad modems, but in my experience these types of things tend to happen due to IOS bugs. Usually a quick fix would be just to reboot the box. Sometimes configuring modem recovery helps this as well. Something like this:
> 
> modem recovery maintenance window 90
> modem recovery maintenance action drop-call
> modem recovery maintenance time 2:00
> modem recovery maintenance stop-time 4:00
> modem recovery maintenance max-download 5
> modem recovery threshold 10
> 
> 
> This will reload the firmware on the modems which can help prevent them from "going bad". I still don't understand why the MICA modems on Cisco boxes have problems like this. We used to use Livingston/Lucent Portmasters prior to switching to Cisco and we never heard of such a thing as a modem going bad or needing firmware reloaded. I think the premise of how modems in those competing boxes worked was entirely different though. I think everything in the Portmaster was software vs. the firmware loading on the AS5xx0's... but whatever.

Hi, Vinny.  Thanks for the reply.  Yes, they are MICA modems.  The IOS 
rev is 12.3(22) running the Enterprise Plus w/ crypto featureset.  I 
bumped it up this spring during the migration.

I implemented the recovery suggestions above.  I'm going to have to do a 
little research to see exactly what each option does but I trust you.

After I posted to the list we got the backup 5300 online.  It has half 
the interfaces and modem capacity but it should suffice.  We 
reprovisioned the 2 circuits that were already connected to it and had a 
tech in that area move patchcords for another pair down to that 5300. 
The downside of this is that it only has 32MB of RAM.  Once it started 
answering calls it quickly ran out of RAM.  I can no longer sh run on 
the 5300.  I managed to get it to about 4MB of free RAM by adding some 
static routes and killing OSPF.  That kept it from crashing in the night.

I've seen this happen before on a 3660 with over 1k ATM PVCs.  When the 
3660 ran low on RAM 'sh run' started skipping large quantities of config 
lines.  Then 'sh run' stopped running altogether.  When I wrote the 
config before rebooting the 3660 it was missing about 2/3s of the 
config.  That was a fun night.

I rebooted the problematic 5300 early this AM and had a CO tech re-add 4 
of the PRIs to the pool.  It is now answering calls with 100% success. 
We'll see if it lasts.

Thanks to everyone for the help.  Hopefully we can keep these 5300s 
running long enough to see us through to the end of our dialup offering...

Thanks
  Justin