Group alerting is currently not implemented. We've already documented this feature request here:
Its an oldie but a goodie :)
Not sure when we would get this done. We'd welcome anyone in the community who would be willing to try to implement it.
I have a question for the community on this. Your responses may help us determine how to implement this.
Suppose I have a group with 3 resources in that group - resource A, B and C.
I apply an alert template to that group. So all 3 resources have the alert.
1) I add a new resource D to the group
Does D automatically get that alert template applied? Or does D not get the alert?
2) I remove resource A from the group
Does A automatically get that alert template removed? Or does A keep that alert still?
In other words, is the alert templates automatically applied/removed as the group membership changes?
Or, do you apply the alerts to the group membership only at the time you add the alert? i.e. the members of the group keep that alert defined, no matter if their association with that group is removed (and any new resources that get added to the group will not get that alert).
My 2c... Logically the alert definition is associated with the group not its individual members, so the alert should always be applied to the current members of the group.
We start with a group:
* You can apply measurement schedule updates - those updates persist to the resources even after you remove them from the group
* You can apply changes to the connection properties - those updates persist to the resources even after you remove them from the group
* You can apply changes to the configuration - those updates persist to the resources even after you remove them from the group
So, to me, a simple implementation of group alerts would be to create the alert definition across all resources in the group, and then allow those alert definitions to remain on the resource even if it's removed from the group.
However, I understand what others are saying. From a user perspective, it looks like you're creating the alert definition on the group itself, and so they should only expect the resources to have those alert definition as long as they remain members of the group.
Really, I can go either way. However, maybe we can have the best of both worlds. Perhaps just have an option on the group alert definition that allows people to flip between these two strategies.
Both of the above solutions only deal with adding convenience to the user. Instead of the user having to do to N resources and create the same alert definition, they out those resources in a group and just create the alert definition once.
But what about conditions like "if 3 out of 5 of the resources in the group go down" or "if 60% of the CPUs in the group are loaded > %80"? This is where the convenience ends and the new features really begins. This would give the user some intelligent control over resources on the group level.
In the end, we'll probably want to implement both of these.
Cinaed, which of these options were you thinking about when you started the thread?