A logistics engineer adds a VLAN on one leaf from the CLI. That VLAN is not part of the fabric intent stored on Nexus Dashboard. At the next deploy from the dashboard, the VLAN is gone. Why?
Select an answer to reveal the explanation.
Short Explanation
You taped a note to a desk that the floor plan does not include, and the next reset of the floor plan took the note with it. Nexus Dashboard deploys the intent it has. A VLAN that exists only in the CLI is not in that intent, so the deploy removes it.
Full Explanation
Nexus Dashboard is the place you express fabric-wide intent, and a deploy pushes that intended configuration to the switches. The logistics VLAN was never added to the intent, so the deployed configuration has no such VLAN and the manual one disappears. The dashboard does not sit in the forwarding path, and it does not drop frames for an unknown VLAN. It removes the configuration object. POAP runs at boot when startup configuration is missing. A deploy against a switch that already has a startup configuration is not POAP, and the presence of a startup configuration is what keeps POAP from running. The guest shell is an on-box Linux environment with its own network namespace. It does not rewrite the VLAN database because a fabric deploy ran. Caveat: if you need the VLAN, add it to the intent and deploy again. Putting it back with the CLI alone schedules the same removal. Operational check: compare the leaf VLAN list with the fabric intent after the deploy and confirm only the intended VLANs remain.