MGCP requested event vs. detect event

Hello,

What is the difference between requested event and detect event in MGCP?

When is an event Quarantined?

Some example would be of great help.

Thanks,

Jack.

 

Re: MGCP requested event vs. detect event

Jack,
A requested event would be sent in an RQNT message, and might be for
something
like "offhook" or "onhook". Once the call agent has sent this to the
gateway,
then when it detects such an event, it will send it via a NTFY message.


If I recall correctly, I think quarantine works like this:
If multiple events occur on an interface, they must be sent one at a
time, with
only one event outstanding in the transmit queue. Let's say the person
presses a series of digits, and the call agent asked for one digit at a
time (not GW digit collection).

Since MGCP sends over UDP, it can't just send them all up to the call
agent
because some messages could be lost, and would require retransmission.
The GW can only send one at a time, in FIFO order. It sends the event,
waits for an ack,
then sends the next event. Otherwise, some UDP messages would get
through,
and others might not. Then the CA has no idea what the real digit order
was supposed
to have been, even though with retransmission it got all of them.

The first digit will be sent to the CA in an NTFY, and the GW waits for
an ACK.
Subsequent events (while an ACK is outstanding) will go into the
quarantine buffer, and dequeued and sent as
events are acknowledged. Same thing goes for something like
offhook/onhook
sequences. These have to be naintained and processed in order, or the
MGC
doesn't know what the current hookstate is.

Hope this is helpful.