CI Lifecycle Management

Definitive questions

• +What is CMDB attestation?
• +What is an orphan CI?
• +When should a CI be marked 'Absent'?
• +What triggers first_discovered?

Scenario drills

1. Stale CIs polluting reports

   “Reports show 4,000 servers but only 3,100 are actually running. How do you clean up without breaking history?”

   Approach

   • Create a Data Manager policy: install_status → Absent after 30 days of no discovery.
   • Escalate: Absent → Retired after 60 additional days if no attestation.
   • Never bulk-delete — you'd lose incident/change history.

   policy.condition = 'last_discovered<javascript:gs.daysAgoStart(30)';
   policy.action    = 'setInstallStatus=Absent';

   Alternate approach

   Alternate: run 'discovery status' review with CI owners quarterly and let humans attest rather than automating retirement for regulated CIs.

   Pitfall

   Auto-retiring CIs that just failed one Discovery cycle causes false 'gone' alarms and breaks impact analysis.