The platform is up, but one dependency path is not.

Use this when failover completes and traffic continues to reach the old node because adjacent hosts never learn the new owner fast enough. These issues are easy to misread because the operating system still responds while one control path, runtime dependency, or trust assumption is already out of alignment.

The symptom usually appears downstream from the real change.

• Linux assumptions are often changed indirectly by updates, policy, hardware changes, or cleanup work that looked harmless at the time.
• keepalived assumptions are often changed indirectly by updates, policy, hardware changes, or cleanup work that looked harmless at the time.
• ARP assumptions are often changed indirectly by updates, policy, hardware changes, or cleanup work that looked harmless at the time.
• Networking assumptions are often changed indirectly by updates, policy, hardware changes, or cleanup work that looked harmless at the time.