I love the concept of the vantage point. Theoretically this should
prevent excess notifications, but in practice it only partially
works. It seems that it only works if the "down" router is probed
first before it probes the devices beyond it (away from the vantage
point). If the devices beyond the router are probed first I get
notifications from all of them up until the router is probed and
found "down." So I'll often get some of the excess notifications, but
not all of them.
A possible solution (optionally enabled by the map design) would be
to have InterMapper back-check the path from any "down" device and
make sure everything between it and the vantage point is "up" before
sending the notification. That should prevent any excess
notifications. Or is there already some way to do this?
However, it's not just the excess notifications that concern me; It's
the missing notifications. If a router goes down for a few minutes,
and while it's down a device behind the router also goes down, and
then the router comes back online, the only notifications I get are
for the router going down and back up. If I don't check the map
manually I'll never know another device went down beyond that router.
Shouldn't I get notifications for devices beyond that router that are
down once the router comes back up (of course only after InterMapper
has a chance to re-probe them)?
This happened to me last night. I saw two notifications, one saying a
router went down (and thank goodness I didn't get the 50+
notifications for the devices beyond it) and another saying it was
back up again a few minutes later. No other indication of trouble,
but a check of the map this morning shows a device beyond the router
went down and I was never notified. I'm scared I'm going to have to
remove the vantage point and sift through all those excess
notifications or always have to check the map immediately any time a
device goes down and back up to make sure everything beyond it was up
as well. :-/
Don't get me wrong, I will check the map eventually, but it's a
question of whether or not I have to cut lunch short every time I see
a device miss a polling cycle and come back 2 minutes later. There
may be something else down beyond that device and I wouldn't know.
Another problem that results from the "excess notification" problem:
I have different people being notified for different devices. For
example, a web server at one location will notify someone on-site if
it goes down. If that web server is beyond a router that goes down,
and the web server happens to get probed first, the web server's
"down" notification is sent to that contact person, but then the
router is probed and seen as "down" thus disabling notifications for
devices beyond it. So when the router comes back on-line the person
who got the "down" notification on the web server never sees a
corresponding "up" notification. He's still under the impression his
web server is down and won't know otherwise until he checks and see
it's "up" and wonders why he never got notified. :-/
I realize that "back tracking" a device before sending a notification
may slow down that notification, and not everyone would want it, but
if it were available I'd definitely use it. However, if that's not
possible, or practical, for some reason, can we at least get
"balance" on the notifications? If a device sends out a "down"
message, set some flag and make sure the "up" notification is sent
when it comes back on-line. After all, if I'm going to get the "down"
message anyway you might as well send the up message as well! And of
course make sure I get the "down" message for any device beyond the
router that doesn't come "up" after the router comes back up. :)
Thanks,
Jay
____________________________________________________________________
List archives:
http://www.mail-archive.com/intermapper-talk%40list.dartware.com/
To unsubscribe: send email to: [EMAIL PROTECTED]
