Extreme Networks ExtremeWare Version 7.8, Troubleshooting Manual

Page: 20 / 122

20 Advanced System Diagnostics and Troubleshooting Guide
“i” Series Switch Hardware Architecture
BlackDiamond MSM Redundancy
The CPU subsystems on a pair of BlackDiamond MSMs operate in a master-slave relationship.
(See
Figure 5 .)
Figure 5: BlackDiamond MSM redundancy scheme
The master MSM CPU subsystem actively manages the switch and the task of switching packets in the
CPU control (or management) path. The slave MSM CPU subsystem is in standby mode, but is checked
periodically by the master MSM CPU (via EDP) to determine whether it is still available.
The master MSM also guarantees that management operations occur in a synchronized manner. For
example, if you make a configuration change and need to save it, the master MSM ensures that the
configuration is saved to both the master MSM and the slave MSM at the same time. That way, the
updated slave is ready to take over as the master if the master MSM fails.
Despite the master-slave switch management role, the switch fabrics on both MSMs actively switch core
traffic in a load-shared fashion. Load-sharing switches core traffic from different I/O modules.
All MSMs share the control (or management) bus on the backplane, and use it to communicate to each
other and to installed I/O modules to perform system health checks and status polling.
Causes of MSM Failover and System Behavior
A number of events can cause an MSM failover to occur, including:
• Software exception; system watchdog timer expiry
•
Diagnostic failure (extended diagnostics, transceiver check/scan, FDB scan failure/remap)
• Hot removal of the master MSM or hard-reset of the master MSM
MSM64i (A)
MSM64i (B)
Switching
Fabric
Fault Tolerant
Switch Fabric
and
System
Management
I /O Module
I /O Module
I /O Module
I /O Module
I /O Module
Backplane
I /O Module
I /O Module
I /O Module
CPU
Sub-
system
Switching
Fabric
CPU Control Path
Data Path
CPU
Sub-
system
DN_028A