📊 Deploy Grafana Dashboards

Use deploy_grafana_dashboards.yml to update provisioned dashboard files on an existing Grafana installation. This path does not install Grafana or Azure CLI, register/rotate Entra credentials, change Grafana authentication, update data sources, or manage the PostgreSQL database.

📋 Prerequisites

• Run from the repository root using the normal Ansible environment and Grafana inventory credentials.
• Grafana must already be installed/configured with its service account, service, and provisioning path. The task checks for the existing grafana_setup_config_file before making dashboard changes.
• The dashboards’ referenced data sources must already exist. In particular, the PostgreSQL Status dashboard references the Prometheus data source with UID prometheus.
• Confirm the inventory’s dashboard/provider paths refer to the running installation. This path does not rewrite Grafana’s main configuration to point at new provisioning directories.

🔎 Preview

ansible-playbook -i inventory/grafana/inventory.ini playbooks/grafana/deploy_grafana_dashboards.yml --syntax-check
ansible-playbook -i inventory/grafana/inventory.ini playbooks/grafana/deploy_grafana_dashboards.yml --list-tasks
ansible-playbook -i inventory/grafana/inventory.ini playbooks/grafana/deploy_grafana_dashboards.yml --check --diff

Add the usual vault password/identity, SSH, or become options if required. Use --limit grafana-0 to select that existing host when appropriate. The path updates the complete repository-managed dashboard set, not only the PostgreSQL dashboard.

▶️ Deploy

ansible-playbook -i inventory/grafana/inventory.ini playbooks/grafana/deploy_grafana_dashboards.yml

The shared dashboard tasks create the dashboard provisioning directories, render the file provider, and install the same dashboard set used by the full Grafana deployment. They also remove the explicitly retired proxmox-vm-snapshots.dashboard.json file, matching the existing full-deployment cleanup; unrelated dashboard files are not pruned.

Dashboard JSON changes are picked up by Grafana’s file provider without a service restart. The configured grafana_setup_dashboard_update_interval_seconds defaults to 30 seconds. If the provider definition itself changes, the existing Restart Grafana handler runs at the end of the play so Grafana loads the new provider settings; plan for a brief interruption in that case.

✅ Verify and Troubleshoot

• Open the configured dashboard folder (default Infrastructure) and refresh the relevant dashboard after the provider interval.
• For PostgreSQL Status, check the PostgreSQL Host Memory Usage graph. It reports database host memory, not PostgreSQL-only process memory; it requires Node Exporter metrics with instance labels matching the PostgreSQL Exporter targets.
• If a dashboard is missing, verify its destination file, provider path, permissions, and Grafana provisioning logs. If a panel shows No Data, inspect the data source and Prometheus query rather than reinstalling Grafana or rerunning Entra registration.
• If the existing-configuration assertion fails, inspect the configured path or perform the separately reviewed full Grafana deployment. Do not bypass the prerequisite merely to create dashboard files on an unconfigured host.
• Roll back dashboard changes through the repository and rerun this dashboard-only path. File-provisioned dashboards may overwrite UI edits according to the provider policy; keep intentional changes in their source JSON files.