IRE Resurrecting Retired CIs During Hardware Refreshes - Anyone Solved This Cleanly?

[Bobby Campbell]

We've got an IRE behavior causing real data quality problems and I'm hoping someone's cracked this already.

The scenario: Infrastructure teams refresh a device and keep the same hostname, though it is new hardware. They retire the old CI, we run Discovery on the new box. Different serial number, so IRE correctly gets no serial match. But then it falls through to name matching, finds the retired CI, reactivates it, and overwrites the serial number. No new CI created. The transition never happened as far as the CMDB is concerned.

I get why the Name fallback exists - serial numbers are unreliable enough that a name match probably does mean "same device, bad serial data" in most cases. But during a hardware refresh, same name + different serial means definitely a different device. The fallback is working against us.

Our current best idea: Add operational_status!=6 as a condition on the name-based identifier entry, so retired CIs can't be matched against. Simple, no process change required from the infra teams.

The problem with that: Our Discovery Admin raised a valid concern - what about CIs that were accidentally retired while the device was still live? With this condition in place, Discovery would just create a duplicate instead of correcting the mistake. We'd have two CIs for the same device with no obvious flag.

What we'd actually prefer is something more like an alert - flag the situation for human review rather than silently blocking or silently resurrecting. Better to catch the problem before the damage is done than discover it later in an audit.

Has anyone implemented something like that? Or found a cleaner solution to the refresh problem altogether? We can't be the only shop where infra teams reuse hostnames.