(Page 6 of 7 in this chapter)

5.5 Handling Faults

The MC1 board is equipped with a watchdog timer. The watchdog timer is used to determine if system software hangs up during normal operation. When enabled, the watchdog timer must be periodically refreshed by the application, or it will electronically disconnect the MC1 board from the MC1 bus. The MC1 board will no longer transmit data or clock signals.

The watchdog timer is refreshed by reading the status of the board (MC1_QUERY_STATUS). If the status of the MC1 interface board is not read by the time the watchdog timer has fired (approximately 9 seconds), the MC1 card will electrically disconnect itself from the MC1 bus.

If a clocking failure is detected on the primary MC1 clock line, the autofallback feature will cause a slave to switch from referencing the primary master clock line to start using the backup clock line.The autofallback applies to slaves only. If a secondary master detects a clocking failure, it reverts to using its internal clock.

You enable the watchdog timer and autofallback feature with the MC1_CONFIG command.

5.5.1 Recovering from Clocking Failures

The application should periodically call the MC1_QUERY_STATUS command to determine when a loss of clock source is detected by the MC1 board and take action to recover.

When the MC1 board configured as primary master fails and the watchdog timer is enabled, an alarm is noted. The MC1 board is automatically disabled from driving the MC1 bus. The secondary master reverts to using its internal clock to drive the backup clock signal. If autofallback is enabled, the slaves automatically switch to reference the backup clock driven by the secondary master. The application should be checking for a failure condition by regularly invoking the MC1 command MC1_QUERY_STATUS.

When a primary master clock failure is detected, the application should do the following:

  1. Demote the primary master board to a slave (if it is still functioning)

  2. Promote the secondary master board or a slave board to be primary master

  3. Promote a slave board to be secondary master (if desired)

  4. Reconfigure the new secondary master board (if there is one) and all slaves to reference the clock driven by the new primary master board

  5. The failed MC1 board can now be powered down and removed from the multi-chassis system by removing the MC1 cable.

When a secondary master clock failure is detected, the application should do the following:

  1. Demote secondary master board to a slave (if possible)

  2. Promote a slave board to secondary master

5.5.2 Network Synchronization Failures

If the network board which is driving the SEC8K clock signal reports a loss of synchronization with the PSTN, choose an alternate network board as the source of synchronization. If the alternate board is in the same chassis as the original network board:

  1. Configure the original network card to stop driving SEC8K

  2. Configure the alternate network board to start driving SEC8K

If the alternate network board is in a different chassis:

  1. Configure the original network board to stop driving SEC8K

  2. Configure the MC1 board in the chassis with the original network board to stop propagating SEC8K to MC1_8K

  3. Configure the alternate network board to start driving SEC8K

  4. Configure an MC1 board in the chassis with the alternate network board to propagate SEC8K to MC1_8K


(Page 6 of 7 in this chapter)

Tech_Support@nmss.com
Copyright © 1996, Natural MicroSystems, Inc. All rights reserved.