Analog Line problem - Only seems to happen on when running on CPU 0

[30n30w]

I've many analog lines hanging off a 61C release 4.5. For the past month or so, a few fire alarm dialers have reported dialing or DACT failures - of course I checked out the analog voltages for on and off hook - no issues, can make calls with a butt-set no problem. Blamed the fire alarm panels, they've been replaced twice.
I've changed feed pairs, put new analog cards in different slots - the two lines are on different cards, different shelves - same issue on both lines to a single fire alarm panel (for this example)
What I have noticed is that the issues only seem to occur when the switch is running on CPU 0. I've got no real errors showing on the switch - I log the MTC and BUG errors and have reviewed them for the last 30 days - nothing significant.
It's been some time since a large system class for me, I think the proper troubleshooting is to start at the endpoint and work back to the core. I'm not having any other issues with this system, just these 4-6 fire alarm panels giving me grief.
The problem exists on more than one loop, so I'm eliminating the SUPL cards, what would be next? the CNI cards ? 3PE? To swap a CNI you have to split the system to keep on side up correct?
As I said, it's been a while - a lot of this knowledge is use-it-or-lose-it and I've lost it.
Thanks for any input.
Andrew

 

[DFKSydney]

To prove the CPU theory, have you stopped the nightly change over? Maybe leave it on each cpu for a week to prove the theory.

Seems to me that unless you do that your fault finding is a bit off.

 

[30n30w]

No, I haven't stopped it because based on the meridian logs and the reports generated by Alarm Central I can tell that running on the CPU 0 side is when things happen. Since I have so many digital and analog lines and most serve a campus 24x7, I don't want to leave it running on a single CPU without redundancy, I don't believe that the CPU is the issue (not showing any errors), but there are cards between the CPU's and the IPE shelves (3PE, CNI) that could be troublesome.
I have spare memory and spare CPU's (hell - I think I have spare everything....) but an old Nortel guy taught me to ask questions if I couldn't remember, but always work from the problem end back towards the core and CPU card. His advice always made a lot of sense because it tends to eliminate chasing ghosts; I just can't remember if the 3PE or CNI comes next in order from the SUPL/Network shelf to the CPU. My notes and books from the class are in storage, and plowing through the pdf versions trying to find what I need isn't getting me very far...It's a little like the Chilton Car Manuals - they tell you EVERYTHING except what you need to know at that moment to fix your car...

Thanks for the input.
Andrew

 

[30n30w]

I think I've determined that the next card in line on the way back toward the CPU is the 3PE. It's the card that allows either CPU to communicate with either network shelf. Now I just need to locate the correct procedure to change it. Since I'm still running the older Motorola processors, a lot of the pdf's I find online are based on CP4 processors and newer hardware; I'm not sure the newer systems even use the 3PE cards anymore....If anyone has a copy of 3.0-4.5 NTP's on CD or knows where they are online, I would appreciate a pointer in the right direction.
Andrew

 

[DFKSydney]

First up, I was not suggesting to split the CPUs, just stop the nightly change over. Redundancy intact.

You say that the failing analogs are in different shelves. Are they in the same network group? Group 0 = loops 0 - 31, group 1 = loops 32 - 63 etc.

If they are in the same network, you may have faulty a TDS/TDR card. LD 34 will let you do the self tests etc. Actually all of these should be tested. This can be done with no user affect.

Do you have FIJI cards?

Basically what I am looking at is going for the "easier fix" and working up, not the other way.

 

[30n30w]

Sorry - I misinterpreted what you had said regarding the switchover. You meant edit the midnight routines, not keep it from switching over. Got it.
This is a 61c so there is only one group - there are 3 DTR cards, and I will test them; however if there is a fault, why only on CPU 0? That's why I hadn't focused on them, they are shared resources common to both sides. I haven't been able to find the specific procedure for changing out a 3PE card, but I *think* it goes like this:

1. make sure system is running on CPU 1 (I'm changing the 3PE on CPU 0 side)
2. disable redundancy/split the system/throw CPU 0 to maint.
3. verify clock is running on 1, disable clock 0
4. disable CNI card on CPU 0
5. Disable 3PE Card on Shelf 0
6. R/R 3PE on Shelf 0
7. Enable 3PE card
8. Reverse 4-1 above to bring back into Redundancy.
9. switchover and see if the issues recur.

Any comments or suggestions on the above procedure?