Goal:

• Accumulation of stale host entries will be an issue in "high-churn" environments (lots of short-lived workloads being enrolled to IPA)
• Insights Inventory application has a "Staleness and Deletion" feature that marks hosts as stale or deletes them after certain (configurable) inactivity thresholds: insights-inventory-staleness-and-deletion.png.
• ipa-hcc-server agent can periodically query Inventory and host-del hosts that:
  • Were enrolled via Podengo (we record the inventory-id)
  • Are no longer present in the Inventory (i.e. have been deleted, whether due to staleness or other action)
  • Are not IPA servers
• If you are having customer conversations, please ask about their churn rate and whether stale host deletion would be important for them.

Acceptance Criteria:

• The feature can be enabled/disabled deployment wide
  • The setting could be controlled from HCC, or from IPA config, or both somehow.  Where the control lives is an implementation detail.
• The initial enable/disable setting is exposed (and can be selected?) in the registration workflow.
• When the feature is enabled, hosts that are removed from Insights Inventory, either due to staleness or deliberate action, get removed from the IPA domain (i.e. their host entry gets deleted)
  • Only hosts that were enrolled via Podengo service get checked and removed.
  • IPA servers are never automatically removed, even if they were enrolled via Podengo and their corresponding Insights Inventory record goes away.