Fleet values sync

Chart fleet-values-sync (charts/all/fleet-values-sync) keeps cross-cluster domain references in the field-content Argo CD Application aligned with ACM inventory — without committing cluster-specific URLs to Git.

It is part of the Validated Patterns clustergroup on hub and spokes. See also GitOps PUSH vs PULL and Region strategy.

What it does

Mode Cluster Action
Hub Hub Reads ManagedCluster east/west + local ingress; patches hub field-content with deployer.domain, clusters.hub|east|west.domain, and API URLs
Hub Hub Pushes fleet-cross-cluster-config ConfigMap to spokes via ACM ManifestWork
Spoke East / West Reads ConfigMap; patches local field-content with clusters.hub.domain and global.hubClusterDomain

What it does not do: import spoke API tokens, create ManagedCluster objects, or patch secrets. Tokens belong in one-time ACM import or RHDP secrets — see RHDP install playbook.

When it runs

  • CronJob fleet-values-sync in namespace openshift-gitops (deployed by clustergroup)
  • Default schedule: every 10 minutes
  • Idempotent: skips patch when helm values already match

Data flow 

flowchart LR
  ACM["ACM ManagedCluster\neast / west"]
  HUB["Hub CronJob\nfleet-values-sync"]
  FC_H["field-content\n(hub)"]
  MW["ManifestWork\nfleet-cross-cluster-config"]
  FC_S["field-content\n(spoke)"]

  ACM --> HUB
  HUB --> FC_H
  HUB --> MW
  MW --> FC_S

Downstream charts consume patched values via clustergroup extraParametersNested — for example clusters.east.domain for Kafka Console, Skupper, Developer Hub topology, and console links.

Prerequisites

  1. Hub field-content Application exists and syncs.
  2. East and west clusters are imported in ACM and show Available.
  3. ManagedCluster status includes API URL (ACM 2.16+: apiserverurl.openshift.io claim).

Until spokes are Available, the job logs ManagedCluster east not found yet and retries on the next CronJob tick.

Verify 

# CronJob deployed
oc get cronjob fleet-values-sync -n openshift-gitops

# Manual run (hub)
oc delete job fleet-values-sync-manual -n openshift-gitops --ignore-not-found
oc create job fleet-values-sync-manual --from=cronjob/fleet-values-sync -n openshift-gitops
oc logs -n openshift-gitops job/fleet-values-sync-manual -f

# Domains in field-content
oc get application field-content -n openshift-gitops -o jsonpath='{.spec.source.helm.valuesObject.clusters}{"\n"}' | python3 -m json.tool

# Spoke received hub domain
oc get configmap fleet-cross-cluster-config -n openshift-gitops -o yaml

Also covered by bash scripts/verify-fleet.sh (checks CronJob presence).

Troubleshooting

Symptom Likely cause Fix
Stale clusters.east.domain on hub Spoke not Available when job last ran Wait for ACM import; run manual job
Spoke missing clusters.hub.domain ManifestWork not applied or spoke CronJob missing Sync fleet-values-sync on spoke; check oc get manifestworks -n open-cluster-management
Wrong apps domain after ACM 2.16 upgrade Old job read kube-apiserver claim only Use chart ≥ fix for apiserverurl.openshift.io (included in this repo)
Kafka Console / Skupper still point at example.com field-content not patched yet Confirm deployer.domain overridden by RHDP or fleet-values-sync; refresh affected Argo apps
You patched tokens into field-content Anti-pattern Remove tokens from GitOps values; use ACM UI import — see troubleshooting