HSP not working after SCPU

[mertech]

I have a Campus redundant CS1000E running 4.5 via a Passport. When both CPU's are rebooted, they seem to come up cleanly, one CPU shows active, the other redundant. The HSP shows up and connected. BUT, when you do a SCPU, both CPU's come back active and unable to 'see' each other. It does not matter which one was active to start with. The hardware has been changed....any ideas folks

 

[johnpoole]

what cpu types? the older 81c cabs have a strap that determines cpu 1 and 0.. if you look at the backplane between slots 10 and 11 you will see a red light on cpu 1 only.. if you do not have just one led on core 1, your missing a strap.. it's located near the led, you can see it if you do not have a third cni

john poole
bellsouth business
columbia,sc

 

[mertech]

the CPU's are CPPIV

 

[sporty2]

Do you have all your patches? I just seen this last week and there was a patch that was not on the DEP list.

sporty2

 

[MrPederman]

This may help with HSP recovery.

The pep says cpp only, but there is a CPP4 version


PRODUCT_LINE: CS1000
============================================

TITLE
=====

CS1000E Duplicate IP Address when ELAN/HSP disconnected on CPP platforms.*** NOTE: SPECIAL HANDLING INSTRUCTIONS. SEE NOTES SECTION FOR DETAILS ***

========================
Release: 4 Issue: 50W
Category: GEN
Superceded By: NONE
Release Date: 20070830
Patch Conflict(s): NONE
Special Instructions: YES
Patch Version: 1
========================

CPP: No auto-recovery after ELAN and HSP recovers. |
|
SETUP FOR THE PROBLEM: |
|
When the ELAN and HSP connections between the two cores of a CPP system are |
disconnected, the inactive core will coldstart and assume the role of the |
active core. So there are two active cores in the system. Both cores are |
using the same active ELAN IP address for the ELAN port (LAN1). In addition, |
the same internal IP address is used for the HSP port (LAN2). This may be |
alright as long as the IP connections are disconnected and the original |
active core has not given up the TDM network control. |
|
However, when the HSP and ELAN connections are recovered and since both |
cores are using the same active ELAN IP address, all IP based applications |
will get confused (duplicate IP address) and the corresponding application |
links (eg, AML, RUDP, ... etc) will not work. |
|
On the call server sides, duplicate IP address message are printed as well. |
|
This condition remains so forever until a manual INI is invoked at one of |
the cores. There is no automatic mechanism to recover from this "duplicate |
IP address" condition. |
|
ACTIONS PERFORMED TO CAUSE THE PROBLEM: |
|
One way to duplicate this problem is to un-plug the ELAN and HSP cables to |
the inactive core. After the inactive core restarts to assume the "active" |
role, re-connect the cables. |
|
EXPECTED RESULTS: |
|
The "duplicate IP address" situation should be detected and recovered |
automatically. |
|
ACTUAL RESULTS: |
|
The system just complains about "duplicate IP address" and remains in that |
condition forever until manual INI is invoked at one of the cores.

Solution:
=========
Restart core 1 with reason 86 when duplicate IP address is detected.


Since the patch contains task based changes, the platform needs to be initialised after patch is in service/ out-of-service if the ELAN/HSP are disconnected.

 

[MrPederman]

I'm sorry, here is the pep number.

Remember to read the special instructions.

Type: General Patch
Release: 04.50W
Number: MPLR23840_CPP4_04.50W
Status: Released
Date: 2007-08-30