Somehow I don't think Dirk and
Andy are on the same wavelength. I may be on yet
another wavelength but let me
see if my summary makes any sense at
all.
Andy had a problem where a
service goes down and since it looks like its going to
take a while to fix, Andy would
like to acknowledge that fact. Said
service doesn't
run all the time, so SA only
checks to see if its up during those times that
its
supposed to be
up.
The suggested way to
acknowledge the lengthy repair time is to put the service
into
maintenance
mode.
Andy points out that if someone
forgets to take the service out of maintenance
mode,
then it doesn't get
checked. Since this service isn't checked all the time, he
suggests
that as a new feature, SA allow
for a self-terminating maintenance interval where
the
termination time is determined
by the check schedule.
Dirk responds
that perhaps the termination time should just be a user specified
value
of either a time of day or an
interval at the time the service is put into maintenance
mode.
He suggests that this might be
more generally useful.
Dirk also indicates that there
is no existing code which can calculate a time in the
future
based on the
check schedule making this option hard to
implement.
Andy insists that his
implementation would be better. Dirk disagrees and here we
are.
As I see the original
problem, maintenance mode is a perfect way to acknowledge that
a
service is being worked on and
I also see Andy's point that maintenance mode could
be
left on by accident preventing
SA from every checking the service again.
Dirk's addition of "maintenance
until" is an elegant solution.
At the time the service is
determined to be down, an operator acknowledges the
problem
and provides and estimated time
to repair with a "maintenance until". At the end of
this
interval, SA
will again monitor the service on the specified schedule, and report
it as down
if "now" is within
the check schedule.
This appears to me to work for
any service being monitored. Those monitored
constantly
and those that are only
checked periodically. Now its time for Dirk and Andy to gang
up
on me because I don't
understand anything! <G>
Regards,
Brad Morgan
IT Manager
Horizon Interactive Inc.
IT Manager
Horizon Interactive Inc.
