diff --git a/monitoring.yml b/monitoring.yml
index f753ad7c..4400869d 100644
--- a/monitoring.yml
+++ b/monitoring.yml
@@ -1,9 +1,23 @@
 ---
-# Playbook for monitoring
+# Deploy Prometheus
 - hosts: prometheus.adm.crans.org
+  vars:
+    # Prometheus targets.json
+    prometheus_targets:
+      - labels:
+          job: node
+        targets:
+          - localhost:9100
+          - vulcain.adm.crans.org:9100
+          - odlyd.adm.crans.org:9100
+      - labels:
+          job: prometheus
+        targets:
+          - localhost:9090
   roles:
     - prometheus
 
+# Monitor all hosts
 - hosts: all
   roles:
     - prometheus-node
diff --git a/roles/prometheus/handlers/main.yml b/roles/prometheus/handlers/main.yml
new file mode 100644
index 00000000..4214def5
--- /dev/null
+++ b/roles/prometheus/handlers/main.yml
@@ -0,0 +1,5 @@
+---
+- name: Restart Prometheus
+  service:
+    name: prometheus
+    state: restarted
diff --git a/roles/prometheus/tasks/main.yml b/roles/prometheus/tasks/main.yml
index 8ae973e8..f1702b94 100644
--- a/roles/prometheus/tasks/main.yml
+++ b/roles/prometheus/tasks/main.yml
@@ -1,9 +1,20 @@
 ---
-
-- name: Instal Prometheus
+- name: Install Prometheus
   apt:
     update_cache: true
     name: prometheus
   register: apt_result
   retries: 3
   until: apt_result is succeeded
+
+- name: Configure Prometheus
+  template:
+    src: prometheus/prometheus.yml.j2
+    dest: /etc/prometheus/prometheus.yml
+  notify: Restart Prometheus
+
+# We don't need to restart Prometheus when updating nodes
+- name: Configure Prometheus nodes
+  copy:
+    content: "{{ prometheus_targets | to_nice_json }}"
+    dest: /etc/prometheus/targets.json
diff --git a/roles/prometheus/templates/prometheus/prometheus.yml.j2 b/roles/prometheus/templates/prometheus/prometheus.yml.j2
new file mode 100644
index 00000000..e96dedea
--- /dev/null
+++ b/roles/prometheus/templates/prometheus/prometheus.yml.j2
@@ -0,0 +1,27 @@
+# {{ ansible_managed }}
+
+global:
+  scrape_interval:     15s # By default, scrape targets every 15 seconds.
+  evaluation_interval: 15s # By default, scrape targets every 15 seconds.
+  # scrape_timeout is set to the global default (10s).
+
+  # Attach these labels to any time series or alerts when communicating with
+  # external systems (federation, remote storage, Alertmanager).
+  external_labels:
+      monitor: 'example'
+
+# Load and evaluate rules in this file every 'evaluation_interval' seconds.
+rule_files:
+  # - "first.rules"
+  # - "second.rules"
+
+# A scrape configuration containing exactly one endpoint to scrape:
+# Here it's Prometheus itself.
+scrape_configs:
+  - job_name: dummy
+    # This reload dynamically the list of targets
+    # You don't need to restart Prometheus when updating targets.json
+    file_sd_configs:
+      - files:
+        - '/etc/prometheus/targets.json'
+