We are running PRTG version 8. We have Parent/Child dependencies set up where all Child Sensors on a Device will pause once the Ping Sensor (Parent) enters into a DOWN state. This works as expected, however, the Child Sensors don't get paused until the Parent Sensor goes DOWN (again, this is as expected). This being the case, we still receive a large amount of alerts from the Child Sensors until the Parent (Ping) Sensor goes through it's next check. This is the bahavior we'd like to see happen:

  1. Host A goes down.
  2. Memory Sensor is the next Sensor that gets checked.
  3. Memory Sensor does NOT send a DOWN alert, instead it requests the Parent (Ping) Sensor to run.
  4. If the Parent (Ping) Sensor
  • is UP, send an alert for Memory Sensor.
  • is DOWN, send an alert that the Ping Sensor is DOWN and put all Child Sensors into "Paused" mode.

Is this something that's supported in version 8?

Article Comments

Dear Douglas,

the solution in this case is to run the Ping Sensor in an interval at least half as short as the other intervals of the other sensors on the same device. So if all other sensors have a 60 seconds interval or longer, run the Ping Sensor with a interval of 30 seconds.
There still may be situations where it overlays a bit, but your suggestion is not really feasible, as it would mean on EACH scan of EACH sensor the Ping would need to check before (because the sensors can't know if the target is down), and that would basically blow up the whole Probe Scheduler.

best regards.


Oct, 2012 - Permalink

The overlay is what we're trying to get around. Our Ping Sensors are all set to check every 60 seconds while all other Sensors check at 15-minute intervals and longer, so that is set up as you describe.

As far as EACH scan of EACH Sensor, that's not what would be necessary. What I'm asking is for is a Child Sensor to request the Parent Sensor to run ONLY if the Child is unable to connect or enters into a DOWN state.

Thanks


Oct, 2012 - Permalink

Sorry, even that is not planned, and would require quite an overhead. With a latency of at least 15 minutes then on the triggers of the "other" sensors, any overlays should be eliminated as well.


Oct, 2012 - Permalink