Solved: increasing "ignored" count on serial interface

maxnpj · ‎08-19-2010

To get right to it;

We have two routers (in Brazil if it matters) connected via frame relay. The ignored count on each serial interface keep increasing although not at a fast rate. The input errors / CRC errors / frame errors / aborts do not increase on either router. We had the carrier test the circuits on both routers the other day and that really increased the input errors / CRC errors / frame errors / aborts as well as the interafce resets and carrier transitions. (less that 100 input errors went to 533731 / 0 CRC errors went to 125154 / 64 aborts went to 211460 on one of the routers...the other end saw an increase in those same stats as well but not that extreme). Now, focusing on the router with the largest increase in errors, based on framing errors, aborts, carrier transitions and interface resets, I was thinking that the (external) CSU or the cabling between the (external) CSU and the demarc may be the issue.I had to make a choice as to which cable the carrier tech was going to check (demarc ext. or V.35) since we had a time limitation. I chose the demarc extension since those seem more prone to errors than a V.35 cable.

The carrier claims there is nothing wrong with the demarc extention, or smartjack and they swapped CSU's on both sides. Today I look at the interface stats again, and the only thing that has increased (since Monday) in the "ignored" count.

Early on, when I first noticed this, the only problem with the interface stats was the 'ignored" count, and some research led me to the buffers. While the buffers showed some failures, they all showed "0 no memory" which led me to believe that while more buffers were required at times, the router never ran out of memory.

At this point, I'm a bit stumped as to where to look next. Both routers have an increasing "ignored' count, so that leads me to believe it is *not* the routers themselves, but the circuit although even that has separate components. Should I be looking at both routers as one? (Meaning errors on one side will show up on the other?). Go back to the carrier? I'm not sure what to ask them at this point.

I'm happy to post any info from the routers that somebody thinks would be relevant. I didn't post anything now since I don't know what anybody would want to see.

OK...I think that's enough for now.

Thanks

--Brian

paolo bevilacqua · ‎08-19-2010

Perhaps you could remove the extended demarc(s) from the picture.

Perhaps a known good spare router and related equipment directly at the demarc.

If you cannot ensure the correct extended demarc cables and shielding was used, narrowing the link will help rule out unknowns.

Incorrect, "ignores" indicate that router is too busy and could not process packets fast enough. It never indicates a problem with circuit.

View solution in original post

fsebera · ‎08-19-2010

Perhaps you could remove the extended demarc(s) from the picture.

Perhaps a known good spare router and related equipment directly at the demarc.

If you cannot ensure the correct extended demarc cables and shielding was used, narrowing the link will help rule out unknowns.

Hope this helps a little.

Frank

paolo bevilacqua · ‎08-19-2010

Perhaps you could remove the extended demarc(s) from the picture.

Perhaps a known good spare router and related equipment directly at the demarc.

If you cannot ensure the correct extended demarc cables and shielding was used, narrowing the link will help rule out unknowns.

Incorrect, "ignores" indicate that router is too busy and could not process packets fast enough. It never indicates a problem with circuit.

paolo bevilacqua · ‎08-19-2010

You are failing to relay the most important infromation:

Router Models ?

Exact IOS ?

Circuit speed ?

"show proc cpu" ?

maxnpj · ‎08-19-2010

Router#1:

2610XM

12.3(24)

2Mb circuit / 1024 PVC

sh proc cpu
CPU utilization for five seconds: 15%/12%; one minute: 8%; five minutes: 8%

(I didn't think you wanted the entire output of sh proc cpu. I will be happy to post if you do want that)

At the same moment the sh proc cpu was run:

sh int

reliability 255/255, txload 222/255, rxload 54/255

Router#2

2610XM

12.3(24)

2Mb circuit / 1024 PVC

sh proc cpu

CPU utilization for five seconds: 13%/12%; one minute: 8%; five minutes: 7%

At the same moment the sh proc cpu was run:

sh int

reliability 255/255, txload 88/255, rxload 250/255

I did not mention in the first post that I have also been watching the "sh proc cpu" output for a while as well, and what you see in this post is pretty normal. I don't ever see either router get much beyond what you see here.

--Brian

paolo bevilacqua · ‎08-19-2010

Your routers are too slow for a full 2 MBPs circuit, so well loaded as these are.

You may get lucky switching to a different IOS with more optimized code, or cleaning up the config of any unncessary processing, but the basic matter is that you have to upgrade to current technology ISR routers.

maxnpj · ‎08-19-2010

Thanks for the answer. That makes sense since they have increased this circuit over time. One last thing....and it's more of a pick your brain question.....what about the 2610XM is too slow? The processor? the bus speed, even with the maximum amount of RAM, it wouldn't matter? The whole thing is just too old?

I'm asking more from an "educate me" and "future reference" point of view.

Thanks again.

paolo bevilacqua · ‎08-19-2010

It is the processor that is slow (very slow). Maxing RAM and flash doesn't help.

These things were really made to handle a T1 the most. It was a different world back then

Please remember to rate useful posts clicking on the stars below.