Device Discovery in a Multi Poller Environment

mikej19

Hi,

I wanted to ask if there is a better way to discovery devices during schedule scanning?

Let me provide details of my environment and why this question is getting asked.

We have around 4500-5000 devices in our network. We are running 2026.2.1 Observability Self Hosted and have recently enabled Engine Load Balancing with 4 active APEs. I have scheduled network discovery scans built to scan the many mgmt subnets we use and have designated one of the active APEs as the discovery poller. When a current device is moved to a different APE, the discovery report shows the device as "found" instead of "changed". This creates additional work to filter through the "found" list to find the devices not in SW.

How do others manage their scheduled discovery scans while using multiple APEs and with the new Engine Load Balancing feature?

mr.castaneda

Thanks for laying out the details of your setup—that context really helps.

With Engine Load Balancing (ELB) turned on, the platform can automatically move eligible ICMP/SNMP‑polled nodes between Additional Polling Engines in the same HA pool to keep the workload even and polling reliable. That engine reassignment happens independently of where your scheduled discovery job is configured to run.

A couple of important implications for the behavior you’re seeing:

• The discovery wizard still runs from the specific engine you chose in the schedule.
• When ELB later moves an already‑monitored node to a different APE, the next scheduled discovery sees that IP/hostname as already in inventory, so it shows up in the results as “Found / already monitored,” not as “Changed.” Engine assignment on its own isn’t treated as a “configuration change” by the discovery UI today.

Because of that, the best practice I’ve seen work with ELB + scheduled discovery is:

1. Use discovery purely to find new or materially changed devices.
   1. In Discovery Central / the discovery results, filter the results view down to New (or New + Changed) so you don’t have to scroll through all of the “Found” devices that ELB may have moved around.
   2. Treat “Found” as “still in SW, no relevant property change” and focus your manual review only on the objects flagged as new/changed.
2. Let ELB own the poller assignments and manage poller changes outside discovery.
   1. Use Manage Nodes (with the Polling Engine column) or a small custom report/dashboard grouped by polling engine if you want to keep an eye on how ELB is distributing nodes, rather than trying to infer that from discovery results.
   2. If you ever need to pin a specific device to a particular APE (for locality or firewall reasons), exclude that node from ELB on the HA pool and manage its engine manually.
3. If you’d like “engine changed” to show as “Changed” in discovery results, that’s great feedback.
   1. Today, there isn’t a knob to make engine moves show up as “Changed” in the discovery UI—it’s working as designed.
   2. This would make a solid feature request (e.g., “Treat polling engine changes as ‘Changed’ in scheduled discovery results when ELB is enabled”), so Product Management can look at whether that classification can be improved in a future release.

I hope this information is useful.