diff --git a/docs_user/modules/proc_adopting-compute-services-to-the-data-plane.adoc b/docs_user/modules/proc_adopting-compute-services-to-the-data-plane.adoc
index 4368f8181..0d55565ae 100644
--- a/docs_user/modules/proc_adopting-compute-services-to-the-data-plane.adoc
+++ b/docs_user/modules/proc_adopting-compute-services-to-the-data-plane.adoc
@@ -826,6 +826,40 @@ $ for CELL in $(echo $RENAMED_CELLS); do
 > done
 ----
 
+. Clean up any {networking_first_ref} agents that are no longer running:
++
+----
+$ oc exec openstackclient -- openstack network agent list
+----
+.. If any agent in the list shows `XXX` in the `Alive` field, verify the Host and Agent Type. If the function of this agent is no longer required and the agent has been permanently stopped on the {rhos_prev_long} host, delete the agent:
++
+----
+$ oc exec openstackclient -- openstack network agent delete <agent_id>
+----
++
+* Replace `<agent_id>` with the ID of the agent to delete.
++
+[IMPORTANT]
+====
+During {rhos_prev_long} to {rhos_acro} adoption, the `tripleo_cleanup` playbook stops {rhos_prev_long} services on Compute nodes, including `neutron-ovn-metadata-agent` and `ovn-controller`. However, these agents do not unregister themselves from the {networking_service}; they stop sending heartbeats. When the new {rhos_acro} deployment creates containerized agents, these agents might register as new agents with different chassis IDs or hostnames rather than updating the old {rhos_prev_long} agent records. This leaves orphaned "dead" agents in the {networking_service} database that show as "Alive: XXX".
+
+If you encounter OVN Metadata agents or OVN Controller agents in the "XXX" state after adoption, you must delete these stale agents:
+
+----
+$ for AGENT_ID in $(oc exec openstackclient -- openstack network agent list | grep 'neutron-ovn-metadata-agent' | grep 'XXX' | awk '{print $2}'); do
+  echo "Deleting stale metadata agent: ${AGENT_ID}"
+  oc exec openstackclient -- openstack network agent delete ${AGENT_ID}
+done
+
+$ for AGENT_ID in $(oc exec openstackclient -- openstack network agent list | grep 'OVN Controller agent' | grep -v 'Gateway' | grep 'XXX' | awk '{print $2}'); do
+  echo "Deleting stale OVN controller agent: ${AGENT_ID}"
+  oc exec openstackclient -- openstack network agent delete ${AGENT_ID}
+done
+----
+
+This issue is more common in deployments that use site-specific naming patterns and unique OVN chassis IDs that differ from other node sets. However, this issue can affect any {rhos_prev_long} to {rhos_acro} adoption where the new deployment does not match the old hostnames or chassis IDs.
+====
+
 . Verify that the {networking_first_ref} agents are running:
 +
 ----
diff --git a/docs_user/modules/proc_adopting-networker-services-to-the-data-plane.adoc b/docs_user/modules/proc_adopting-networker-services-to-the-data-plane.adoc
index 823606d11..533b5df9f 100644
--- a/docs_user/modules/proc_adopting-networker-services-to-the-data-plane.adoc
+++ b/docs_user/modules/proc_adopting-networker-services-to-the-data-plane.adoc
@@ -367,9 +367,30 @@ $ oc exec openstackclient -- openstack network agent list
 .. If any agent in the list shows `XXX` in the `Alive` field, verify the Host and Agent Type, if the functions of this agent is no longer required, and the agent has been permanently stopped on the {rhos_prev_long} host. Then, delete the agent:
 +
 ----
-$ oc exec openstackclient -- openstack network agent <agent_id>
+$ oc exec openstackclient -- openstack network agent delete <agent_id>
 ----
 * Replace `<agent_id>` with the ID of the agent to delete, for example, `856960f0-5530-46c7-a331-6eadcba362da`.
++
+[IMPORTANT]
+====
+During {rhos_prev_long} to {rhos_acro} adoption, the `tripleo_cleanup` playbook stops {rhos_prev_long} services on Compute nodes, including `neutron-ovn-metadata-agent` and `ovn-controller`. However, these agents do not unregister themselves from {networking_service}; they stop sending heartbeats. When the new {rhos_acro} deployment creates containerized agents, these agents might register as new agents with different chassis IDs or hostnames rather than updating the old {rhos_prev_long} agent records. This leaves orphaned "dead" agents in the {networking_service} database that show as "Alive: XXX".
+
+If you encounter OVN Metadata agents or OVN Controller agents in the "XXX" state after adoption, you must delete these stale agents:
+
+----
+$ for AGENT_ID in $(oc exec openstackclient -- openstack network agent list | grep 'neutron-ovn-metadata-agent' | grep 'XXX' | awk '{print $2}'); do
+  echo "Deleting stale metadata agent: ${AGENT_ID}"
+  oc exec openstackclient -- openstack network agent delete ${AGENT_ID}
+done
+
+$ for AGENT_ID in $(oc exec openstackclient -- openstack network agent list | grep 'OVN Controller agent' | grep -v 'Gateway' | grep 'XXX' | awk '{print $2}'); do
+  echo "Deleting stale OVN controller agent: ${AGENT_ID}"
+  oc exec openstackclient -- openstack network agent delete ${AGENT_ID}
+done
+----
+
+This issue is more common in deployments that use site-specific naming patterns and unique OVN chassis IDs that differ from other node sets. However, this issue can affect any {rhos_prev_long} to {rhos_acro} adoption where the new deployment does not match the old hostnames or chassis IDs.
+====
 
 
 .Verification
diff --git a/tests/roles/dataplane_adoption/tasks/neutron_agents_cleanup.yaml b/tests/roles/dataplane_adoption/tasks/neutron_agents_cleanup.yaml
index 8a2b04032..7b1c979a8 100644
--- a/tests/roles/dataplane_adoption/tasks/neutron_agents_cleanup.yaml
+++ b/tests/roles/dataplane_adoption/tasks/neutron_agents_cleanup.yaml
@@ -25,3 +25,33 @@
       ${BASH_ALIASES[openstack]} network agent delete ${AGENT_ID}
     fi
   loop: "{{ neutron_ovn_controller_gateway_agent_cleanup_hosts | default([]) }}"
+
+- name: delete stale OVN Metadata agents from TripleO
+  ansible.builtin.shell: |
+    {{ shell_header }}
+    {{ oc_header }}
+    alias openstack="oc exec -t openstackclient -- openstack"
+
+    # Delete metadata agents that are not alive (stopped during TripleO cleanup)
+    # This handles cases where the new EDPM deployment creates new agent records
+    # instead of reusing the old TripleO ones (different hostnames/chassis IDs)
+    for AGENT_ID in $(${BASH_ALIASES[openstack]} network agent list | grep 'neutron-ovn-metadata-agent' | grep 'XXX' | awk '{print $2}'); do
+      echo "Deleting stale metadata agent: ${AGENT_ID}"
+      ${BASH_ALIASES[openstack]} network agent delete ${AGENT_ID} || true
+    done
+
+- name: delete stale OVN Controller agents from TripleO
+  ansible.builtin.shell: |
+    {{ shell_header }}
+    {{ oc_header }}
+    alias openstack="oc exec -t openstackclient -- openstack"
+
+    # Delete OVN controller agents that are not alive (stopped during TripleO cleanup)
+    # This handles cases where the new EDPM deployment creates new agent records
+    # instead of reusing the old TripleO ones (different hostnames/chassis IDs)
+    # Note: grep -v 'Gateway' excludes OVN Controller Gateway agents which are
+    # handled by the dedicated task above
+    for AGENT_ID in $(${BASH_ALIASES[openstack]} network agent list | grep 'OVN Controller agent' | grep -v 'Gateway' | grep 'XXX' | awk '{print $2}'); do
+      echo "Deleting stale OVN controller agent: ${AGENT_ID}"
+      ${BASH_ALIASES[openstack]} network agent delete ${AGENT_ID} || true
+    done