Alcatel Carrier Internetworking Solutions 6648 - Manual (Page 47)

Troubleshooting the Switch System Troubleshooting IPC on OS-6/7/8XXX Series of Switches
OmniSwitch Troubleshooting Guide September 2005 page 1-27
Troubleshooting IPC on OS-6/7/8XXX Series of
Switches
IPC (Inter Process Communication) is should by the system to communicate between different software
modules. This communication can be between different processes in the same software module or between
two entirely separate modules. This process can be between NI and CMM or between CMM to CMM.
Burst Bus commonly known as BBUS (management bus) is used for the IPC communication. IPC uses
connectionless build-in Vx Works sockets to communicate.
Typical problems that can arise because of the problems with IPC can cause the following symptoms:
• Loss of access to the console of the switch
• Loss of messages between CMM and NI resulting in switching and routing problems.
• High CPU utilization on CMM
Debugging IPC
IPC has 5 different buffer pools:
• Urgent Pools
• Control Pools for control messages
• Normal Pools for some control messages as well as other messages
• Jumbo Pools
• Local Pools
Each of these pools have some dedicated buffers available. Once any of these processes initiates a socket
to communicate, it is suppose to tear the socket down after the communication is done. If it does not tear
the socket then it might result in occupying the buffer space which will not be available for other
processes.
IPC pools can be looked in dshell using the command:
Working: [Kernel]->ipc_pools
UrgentPool: Full size is 1024, remaining: 1024
In socket queues: 0 Not queued: 0:
In DMA queues: 0
ControlPool: Full size is 5096, remaining: 5090
In socket queues: 1 Not queued: 3:
In DMA queues: 2
NormalPool: Full size is 2024, remaining: 2022
In socket queues: 0 Not queued: 2:
In DMA queues: 0
JumboPool: Full size is 256, remaining: 255
In socket queues: 1 Not queued: 0:
In DMA queues: 0