I want to use checkmk to monitor a set of event infrastructure devices that is being used on a few different events. The challenge:
- there’s a core set of things that is being used at all events (some servers, core switches etc. - the basic infrastructure that is needed in each case)
- there are additional devices whose usage depend on the event:
- event A uses switches 1 through 8, but might sometimes also need a few more
- event B uses switches 1 through 4 and a few POS terminals
- event C uses a different number of these switches each time
I’m thinking on how to make this easily manageable, like “presets” or “templates”. My goal here is to be able to enable an “event A” or “event B” template etc., depending on which event you are, in order to avoid having to set/unset criticality: not monitored on a bunch of devices each time you build up one of these events. This will have the advantage that during/after buildup, staff doesn’t have to work checklists and test everything, they can (optimally) just check the monitoring to see if everything is wired and set up correctly.
The monitoring instance itself will be the CRE/community edition running on a VM running on the event infrastructure.
It can’t fully be templated because the number of switches can vary a bit, so a little manual labour will be inevitable. But for the rest, I’m thinking about what would be the best way to structure things.
I thought I could use Folders to separate devices into categories like switches, POS terminals, servers and so on, place the devices that are always present in these folders, and for “optional” devices I could make subfolders (so I can enable/disable all hosts in there by changing the folder properties).
However, if event A has switches 1..10, event B has 1..4 and event C has probably even less, I can’t really make a subfolder per event because I can’t place switch 1 in multiple subfolders. Tags/labels would probably be a better fit for that…
I also thought about just configuring everything once per event and then making a snapshot of the VM, and then the event tech can just rollback to the snapshot of event X, and should something change, they’d just make a new shapshot. But that would mean there would be no historical data, so that’s not an optimal solution.
Does anyone here have a clever idea for this - admittedly special and unique - requirement?
TIA!