Advanced Alerts and Groups

Question

Ive been browsing around for a while and just decided to post.

Maybe im missing something (it would not be surprising) but i havent been able to come up with the following really simple solution.

I have a situation where I need to know when all of a group of nodes are available and when all of them are unavailable.

What i dont want to do is add the nodes individually to a container. So all the nodes have a matching custom property to identify them as a whole. How do i define the alert to trigger and reset based on the status of the group?

Any thoughts? =)

scottd · Answer

Using advanced alerting will be a little clunky, but  it can get you close to what you are asking for, or it may be good enough to meet your requirements, not sure...

By using suppression you can make the alert not fire until all of the nodes in the group are not responding, however;

Pitfall #1 -  alert manager will send out an alert for each node in the group once they have all reported 'not up', so if there are 10 nodes in the group, you would receive 10 node down alerts at the same time.

Pitfall #2 - Because the alerts are treated as individuals, they will clear individually as the nodes come back to life, not as a group.

The example below does what is described above...

Example - a trigger condition could look like;

Node Status is not equal to UP

CustomProperty is equal to SWtest

the alert supression would look like;

Node Status is equal to UP

CustomProperty is equal to SWtest

FormerMember · Answer

Thanks for the feedback. I was hoping there might be a way to accomplish this through advanced alerting with NPM. If there are no other options i can consider this though.

anachronism · Answer

If you have APM you could do this by a SQL query monitor on your SolarWinds database, something like:

select count(NodeID) from Nodes where Status=1 and  DeviceGroup='CORE';

assuming you have a custom property called DeviceGroup, with some devices set to have a value of CORE, this will return the number of devices with status=1 (UP). If this query returns the value of 0, either there are no devices in that group, or all the devices that are in the group are not up (although they may not be down, they may be unkown.... but you probably want to investigate that anyway 
 ).

It's a bit of an ugly hack, and there may be a better solution, but that should work (if you have APM).