Problem linking NM-1A-E3's back to back

tom.storey · ‎07-26-2010

Hi All,

Apologies if this is not the right forum, I couldn't find one for hardware (if this is indeed a hardware fault...)

I have a number of 2811 and 3745 routers which I am trying to link back to back (at this stage) in my lab for staging, they will eventually be deployed and connected via E3 microwave links.

I am having a problem in that certain pars of the modules will work perfectly well together, but other pairs wont.

A brief overview of the topology:

2811 with 1 x E3 module [A] <--> 3745 with 2 x E3 modules [B, C] <--> 3745 with 2 x E3 module [D, E] <-> eventually another 2811 with 1 x E3 module [F]

E3 modules A and B will link fine, as will C and D, and also A and E. I have not yet gotten around to testing E and F yet...

But if I try to connect A and D, module A will have an OOF indicator which blinks on and off irregularly (and on the console I see the interface continuously going up/down), and D will have constant FERF and be down/down. Likewise if I link A and C.

If I link B and C, B will have constant OOF and C will have constant FERF.

Try any combination you like other than A-B and C-D and you get this type of behaviour.

The modules are second hand, and I could try to have them replaced, but is it possible that all 4 would be dodgy? And what in particular about A-B, C-D lets them work, but not other combinations?

My config is quite simple on each interface:

interface ATMX/0

no ip address

no atm ilmi-keepalive

!

interface ATMX/0.1 point-to-point

ip address 192.168.0.x 255.255.255.252

pvc 0/1

cbr 34000

!

and thats it. Just change the interface name and IP to suit each end of the link. I have tried different cables, so I am pretty confident its not a physical layer issue (unless all of my cables are dodgy, but that still doesn't explain why certain combinations work and others don't).

Can anyone suggest any commands I can run, debugs I can turn on and the likes to try and work out whats going on? Are there any tricks to operating two of the same type of card in a 3745, like particular slots they should be installed in (one would hope not)?

I dont feel comfortable rolling these things out into the field, if we have issues and I cant guarantee that any combination of card will work its inevitably going to lead to frustration and more down time trying to shuffle things around to get it working again.

Thanks.

Tom

tom.storey · ‎07-26-2010

Ok a little bit of digging around and Ive found some debug commands, I am now seeing this (output from the 2811):

*Jul 27 13:06:53.999 ACST: clock change removed for int ATM1/0

*Jul 27 13:06:54.999 ACST: clock change done for int ATM1/0

*Jul 27 13:07:03.999 ACST: clock change removed for int ATM1/0

*Jul 27 13:07:04.999 ACST: clock change done for int ATM1/0

*Jul 27 13:07:09.999 ACST: clock change removed for int ATM1/0

*Jul 27 13:07:10.999 ACST: clock change done for int ATM1/0

and so on.

This is with "atm clock INTERNAL" set on the 3745 (interface D).

If I remove "atm clock INTERNAL" from interface D and put it on interface A I get this:

*Jul 27 13:10:10.239 ACST: %LINEPROTO-5-UPDOWN: Line protocol on Interface ATM1/0, changed state to up

*Jul 27 13:10:12.251 ACST: %LINEPROTO-5-UPDOWN: Line protocol on Interface ATM1/0, changed state to down

*Jul 27 13:10:16.251 ACST: %LINEPROTO-5-UPDOWN: Line protocol on Interface ATM1/0, changed state to up

*Jul 27 13:10:18.251 ACST: %LINEPROTO-5-UPDOWN: Line protocol on Interface ATM1/0, changed state to down

and so on.

With a few debug commands on ("debug atm errors", "debug atm events", "debug atm sig-error", "debug atm sig-events") I see the following, but none of it looks "bad" per se, just telling me in more detail that the interface is going up/down:

*Jul 27 13:11:25.307 ACST: ATM(ATM1/0): ATM not active yet

*Jul 27 13:11:25.315 ACST: %LINK-3-UPDOWN: Interface ATM1/0, changed state to up

*Jul 27 13:11:25.315 ACST: atmoc3_cstate(ATM1/0): state=4 atm_lineup=0

*Jul 27 13:11:25.315 ACST: Resetting ATM1/0

*Jul 27 13:11:25.315 ACST: ATMSIG-IF:atmSig_clearAllSvc: Port ID : ATM1/0 Interface Reset

*Jul 27 13:11:27.315 ACST: %LINK-3-UPDOWN: Interface ATM1/0, changed state to down

*Jul 27 13:11:27.315 ACST: atmoc3_cstate(ATM1/0): state=0 atm_lineup=0

*Jul 27 13:11:27.315 ACST: Resetting ATM1/0

*Jul 27 13:11:27.315 ACST: ATMSIG-IF:atmSig_clearAllSvc: Port ID : ATM1/0 Interface Reset

*Jul 27 13:11:31.319 ACST: %LINK-3-UPDOWN: Interface ATM1/0, changed state to up

*Jul 27 13:11:31.319 ACST: atmoc3_cstate(ATM1/0): state=0 atm_lineup=0

*Jul 27 13:11:31.319 ACST: Resetting ATM1/0

*Jul 27 13:11:31.319 ACST: ATMSIG-IF:atmSig_clearAllSvc: Port ID : ATM1/0 Interface Reset

*Jul 27 13:11:33.319 ACST: %LINK-3-UPDOWN: Interface ATM1/0, changed state to down

*Jul 27 13:11:33.319 ACST: atmoc3_cstate(ATM1/0): state=0 atm_lineup=0

*Jul 27 13:11:33.319 ACST: Resetting ATM1/0

*Jul 27 13:11:33.319 ACST: ATMSIG-IF:atmSig_clearAllSvc: Port ID : ATM1/0 Interface Reset

*Jul 27 13:11:45.307 ACST: ATM(ATM1/0): ATM not active yet

*Jul 27 13:11:49.343 ACST: %LINK-3-UPDOWN: Interface ATM1/0, changed state to up

Any assistance would certainly be appreciated.

Thanks,

Tom

tom.storey · ‎07-29-2010

Bump.

Anyone?

Kevin Brennan · ‎07-30-2010

Hi Tom,

I've never worked with these interface modules, but is it possible that you have to set a clock on one side of the back to back circuit? (In the real world, this clock would be provided by your mw cpe)

A quick search turns up "atm clock internal" under the physical link - maybe try that on one side of each link?

HTH

Kevin

tom.storey · ‎07-30-2010

Hi Kevin,

Yep Ive got "atm clock internal" set on one end of each link, but it hasn't helped.

Thanks,

Tom

Kevin Brennan · ‎07-30-2010

That's me out of ideas then!

Kevin

Giuseppe Larosa · ‎07-30-2010

Hello Tom,

one side needs to provide clock to the other as in any back to back setup

note also that you cannot use vpi 0 vci 1 this is reserved for the ATM network

first usable vpi/vci combination is 0/32 , you can use 1/1 but not 0/1

Hope to help

Giuseppe

tom.storey · ‎07-30-2010

Hi Giuseppe,

Thanks for that pointer (you can tell Im not an ATM guru). I wonder if that is causing these issues then, I will have to try it out with PVC 0/32 or 1/1...

Thanks!

tom.storey · ‎08-01-2010

Ok well the VPI/VCI combo wasnt the issue, perhaps unsurprisingly.

Currently setting up a mock radio link in our lab and will test the gear through there, maybe, just maybe, everything will work fine through a radio link and this wont be a problem in the real world.