After reconfiguring or reinstalling the Regional Harbor supervisor service in VCF Automation (VCFA), the Harbor entry on the Services page becomes stuck and shows a status of Unknown. One or more of the following symptoms may be present:
This issue can occur regardless of whether the customer environment is following supported or unsupported configuration paths (see Cause)
VCF Automation 9.1.0.0
Regional Harbor currently supports only the default Nginx load balancer configuration. Using Ingress or a Contour HTTP proxy in front of Regional Harbor is not a supported configuration in this release.
When Harbor's configuration values are edited a second time before the first edit finishes reconciling, the merged configuration can end up with the same setting defined twice (a duplicate key) in the underlying values document. This malformed configuration fails validation, and the service's reconciliation process pauses and does not automatically retry or recover — even if the service is later uninstalled from VCF Automation. Because the service is left in this stuck, unresolved state, the VCFA Services page has no valid status to display and falls back to showing Unknown.
Do not configure Regional Harbor with Ingress or Contour HTTP proxy settings; use the default Nginx load balancer configuration only.
Because the service is stuck in Unknown status, its normal UI actions (Uninstall, Reinstall, Reconfigure) are unavailable or do not complete, so the service must instead be uninstalled and reset via the API. Try the following steps in order.
Common prerequisites for both steps:
Step 1: Uninstall the service via API
Attempt to uninstall the stuck service by setting its lifecycle state to inactive, and clear its stored effective values in the same request so the service is left clean for the next installation:
VCFA_HOST=<vcfa-host>
SERVICE_ID=<service-id>
VCFA_TOKEN=<token>
curl -sS -k \
--request PATCH \
--url "https://${VCFA_HOST}/api/extension/broadcom/service-manager/v2/vcf-services/${SERVICE_ID}" \
--header 'Accept: application/*+json;version=ALPHA' \
--header "Authorization: Bearer ${VCFA_TOKEN}" \
--header 'Content-Type: application/json' \
--data '{"lifecycleState": "inactive", "values": {"effective": ""}}'
Check whether the Harbor entry now returns to a Not installed state on the Services page.
If yes, reinstall Regional Harbor without enabling Ingress or Contour HTTP proxy settings, and the issue is resolved.
If the service remains stuck in Unknown, or the PATCH request fails/has no effect, proceed to Step 2.
Step 2: Force-delete the stuck service via API
If uninstalling via the API in Step 1 does not clear the stuck state, force-delete the service entry so that VCF Automation can rediscover and reinstall it cleanly:
VCFA_HOST=<vcfa-host>
SERVICE_ID=<service-id>
VCFA_TOKEN=<token>
curl -sS -k -w '\nHTTP status: %{http_code}\n' \
--request DELETE \
--url "https://${VCFA_HOST}/api/extension/broadcom/service-manager/v2/vcf-services/${SERVICE_ID}?forceDelete=true" \
--header 'Accept: application/json;version=ALPHA' \
--header "Authorization: Bearer ${VCFA_TOKEN}" \
--header 'Content-Type: application/json'
A 2xx response indicates the delete request was accepted. After that:
If the force-delete does not clear the stuck state, or if both a Harbor RDE and an associated Installation UI custom resource remain stuck, contact Broadcom Support for assistance, as manual cleanup of both objects may be required.