Section 1
SUPERVISOR PROTOCOL
8
Let’s start this example at time slot 1. Machine 1 sends a net message. Machine 2 receives
the message and establishes the time slot as 2. Machine 2 immediately sends it’s net sta-
tus message establishing the slot at 3. Since machine 3 is off line no message will be sent
and after .125 sec. the time slot will rotate to 4. Machine 4 will send it’s net status message,
and the time slot will roll over to slot 0. Since there is no monitor, after .5 sec. the time slot
will rotate to 1 and machine 1 will send it’s net status message. This cycle will repeat until
other machines go off or on line. When a machine is switched on line or powered up it does
not know what the time slot is until it receives a net message from another machine. If only
one machine is put on line, or multiple machines are powered up at the same time, none of
them will know what the current time slot is. To overcome this deadlock the following
scheme is used.
When a machine comes on line it waits for an initial timeout period of 1.5 seconds times it’s
COM ID#. For example if it’s COM ID# is 4 then it will wait 6 seconds. If a message is
received during the initial timeout period then the time slot will be known because the
senders COM ID# is imbedded in the message. For example if a net message from machine
2 is received then the time slot is 2.
If no messages are received within the initial timeout period then a net message is sent. It
can be assumed that either there are no other machines on line or that they have been put
on line at the same time and are waiting for their initial timeout to expire. This keeps sever-
al machines that are powered up at the same time from all sending net status messages
simultaneously.