Forum Discussion

Joe_Tran's avatar
Joe_Tran
Icon for Advisor rankAdvisor
8 years ago

Alert clustering based on matching datasource instances across grouped devices

I have a device group that has the same datasource applied. This datasource auto-discovers and will spin up matching instances across all devices in the group. I would like to have clustered alerts based on the matched instances across all devices in the group.

For example, (pardon the ASCII-like visualization)

ClusterGroup
 |__ Device1
 |     |__ DatasourceA
 |           |_ Instance_ABC
 |               |_ Datapoint_I
 |               |_ Datapoint_II
 |          |_ Instance_DEF
 |                |_ Datapoint_I
 |               |_ Datapoint_II
 |           |_ Instance_GHI
 |               |_ Datapoint_I
 |               |_ Datapoint_II
 |__ Device2
 |    |__ DatasourceA
 |          |_ Instance_ABC
 |               |_ Datapoint_I
 |               |_ Datapoint_II
 |          |_ Instance_DEF
 |               |_ Datapoint_I
 |               |_ Datapoint_II
 |          |_ Instance_GHI
 |               |_ Datapoint_I
 |               |_ Datapoint_II
 |__ Device3
      |__ DatasourceA
            |_ Instance_ABC
                 |_ Datapoint_I
                 |_ Datapoint_II
            |_ Instance_DEF
                 |_ Datapoint_I
                 |_ Datapoint_II
            |_ Instance_GHI
                 |_ Datapoint_I
                 |_ Datapoint_II

If Instance_ABC's Datapoint_I is alerting at the specified cluster threshold in my hypothetical group, I want to generate a cluster alert. If some time afterwards, the situation in my environment gets worse and Instance_GHI's Datapoint_II is alerting at the specified cluster threshold, I want another cluster alert for that instance-datapoint as well.

 

1 Reply

Replies have been turned off for this discussion
  • Bringing this one back up. If I have one resource accessed by multiple devices, and therefore monitored on them, I don't wan't an alert per device. The only option on cluster alerts at the moment would mean looking at all instances, so I can't build a cluster alert per instance as I would require.

    Doing this through service insights would have the same effe