Replace telegraf with prometheus exporters

Still missing SMART and ping, but those can come later.

Swapped as the polling model for prometheus doesn't play especially well with telegraf, and leads to gaps in data

ansible/roles/forrest/files/prometheus/prometheus.yml

@@ -5,14 +5,18 @@ alerting:
             - alertmanager:9093
 
 scrape_configs:
-  - job_name: telegraf
-    metric_relabel_configs:
-      - source_labels: [__name__]
-        regex: go_.+
-        action: drop
+  - job_name: pve
     static_configs:
       - targets:
-          - "{{ pve_hosts.pve.ip }}:9273"
+          - "{{ pve_hosts.pve.ip }}"
+    metrics_path: /pve
+    relabel_configs:
+      - source_labels: [__address__]
+        target_label: __param_target
+      - source_labels: [__param_target]
+        target_label: instance
+      - target_label: __address__
+        replacement: proxmox_exporter:9221
 
   - job_name: traefik
     metric_relabel_configs:
@@ -102,3 +106,13 @@ scrape_configs:
     metrics_path: /projects/{{ healthchecks_project_uuid }}/metrics/{{ healthcheck_api_token }}
     static_configs:
       - targets: [healthchecks.io]
+
+  - job_name: node
+    metrics_path: /metrics
+    static_configs:
+      - targets:
+          - "{{ pve_hosts.pve.ip }}:9100"
+    metric_relabel_configs:
+      - source_labels: [__name__]
+        regex: go_.+
+        action: drop