Allied Telesis SwitchBlade x3100 Series Manual page 437

EPSR Protocol
4.6.4.1 EPSR Protocol Procedures
The master node and the transit nodes that make up the Ethernet ring use the Fault Detection and Fault Resto-
ration procedures provided by the EPSR protocol to maintain the continuous flow of the non-control traffic in
the ring.
Fault Detection Procedure - Two kinds of fault detection procedures are defined to detect a fault. The poll-
1.
ing procedure is the fail-safe mechanism executing in the master node in case the unsolicited fault message
procedure in a transit node fails
Master Node Polling Fault Detection Procedure
•
Transit Node Unsolicited Fault Message Fault Detection Procedure
•
Master Node and Transit Node Fault Detected Correction Procedure
2.
Fault Restoration Procedure
3.
Master Node Restoration Procedure
•
Transit Node Restoration Procedure
•
Each of the above procedure are discussed in the following subsections.
4.6.4.2 Master Node Polling Fault Detection Procedure
The master node uses the polling procedure as a fail-safe mechanism to detect a fault in the ring. It does this by
sending an EPSR health control message via its PP port (only) every HELLOTIME seconds (which is a configured
value as seen earlier). Under normal conditions, when there is no fault in the ring, this health message will make
it across the network and will be received by the master node over its SP port. However, if there is a fault any-
where in the network, the health message will not be received by the master node over its SP port. To detect
this condition, the master node starts a failover timer (using the configured FAILOVERTIME) every time it sends
the health message. If the health message is not received by the master node over its SP port before the failover
timer expires, then it declares a fault in the ring and takes appropriate measure.
Because of the fact that messages could get lost in the network, the FAILOVERTIME configured value
Note:
must be at least twice the value of the HELLOTIME configured value.
4.6.4.3 Transit Node Unsolicited Fault Message Fault Detection Procedure
Unlike the polling procedure described above, where the burden is upon the master node to eventually detect a
fault in the ring in a fail-safe manner, this procedure is used by the transit node to detect a fault on its attached
ring port and immediately notify the master node of the fault. This is accomplished by sending an EPSR Links-
Down control message over a functioning link. A fault link spans two nodes; therefore, both of the transit nodes
that detect the fault send the EPSR Links-Down control message to the master node. When this occurs, the
transit node(s) in question alter the state of the EPSR domain from Links-Up state to Links-Down and maintains
this state until the transit node Fault Restoration procedures are executed. Also, the state of the faulty port is
set to Blocked. However, the state of the functioning ring port is maintained at Forwarding.
Software Reference for SwitchBlade x3100 Series Switches (Layer Two Switching)
Introduction
4-117