Re: Heartbeat stopped messages

[Stuart Booth <lists@xxxxxxxxxxxxxxxx>]

On Tue, 15 Jun 2004 02:57:38 +0000, mcs101main@xxxxxxx wrote:

>In my opinion there should be some mechanism at the protocol level
whereby a node can notify that it is degraded.

--

That would be useful. But there are so many possibilities for
degradation that are application or device specific. A RR might be
unplugged, a mail server might not be responding, a serial port might
have blown up catastrophically....

>  It can be relegated to a lower level class, but it seems that the
periodic heartbeat exists solely for the purpose of knowing the availabilty
of a function.  A function that can respond to a hearbeat, but cannot
perform is primary function should not show up on the network the same as a
fully functional node.

--

The h/b is saying that the device or app is still around rather than
individual functions (subaddresses perhaps?) of the device. It does
seem like a great idea if the 'primary function' cannot be performed,
but if some subaddress of that 'primary function' has failed then
perhaps a heartbeat could include a BSC style body: e.g.

xap-hbeat
{
...
}
Heartbeat
{
01=OK
02=Disconnected
03=Not Responding
04=OK
}

>  A node could also exit when it recognizes that it is degraded, but
then the system would not know that a problem exists and maintenance is
required.

--

It could also send out xAP.Error messages when problems occur. A
diagnostic monitor (such as Viewer in a future update) could track
these, as could a cache/database logger of some kind.

S
--
Stuart Booth <stuart@xxxxxxx>
xAPFramework.net - a xAP software development framework for .net

http://www.xapautomation.org/ 
     http://www.xapframework.net/