Add a script to send queue metrics to CentOS's Zabbix

Fixes: https://pagure.io/fedora-infrastructure/issue/12617

Signed-off-by: Aurélien Bompard <aurelien@bompard.org>

roles/rabbitmq_cluster/files/send-rabbitmq-queue.py

@@ -0,0 +1,81 @@
+#!/usr/bin/env python3
+
+from argparse import ArgumentParser
+from configparser import ConfigParser
+from socket import getfqdn
+from subprocess import run
+from urllib.parse import quote
+
+import requests
+
+
+API_URL = "http://localhost:15672/api"
+CONF_PATH = "/etc/nrpe.d/rabbitmq_args.ini"
+VHOST = "/pubsub"
+METRICS = ("messages", "messages_ready", "messages_unacknowledged")
+
+
+def get_http_client(conf_path):
+    config = ConfigParser()
+    config.read(conf_path)
+    client = requests.Session()
+    client.auth = (config.get("common", "username"), config.get("common", "password"))
+    return client
+
+
+def parse_args():
+    parser = ArgumentParser()
+    parser.add_argument("queue_name", nargs="+")
+    parser.add_argument(
+        "--conf", default=CONF_PATH, help="path to NRPE's rabbitmq_args.ini file"
+    )
+    parser.add_argument("--vhost", default=VHOST, help="the rabbitmq vhost")
+    parser.add_argument(
+        "-s", "--host", default=getfqdn(), help="the server name in Zabbix"
+    )
+    parser.add_argument("--tls-psk-identity", help="passed down to zabbix_sender")
+    parser.add_argument("--tls-psk-file", help="passed down to zabbix_sender")
+    return parser.parse_args()
+
+
+def get_queue_values(http, vhost, queue_name):
+    response = http.get(f"{API_URL}/queues/{vhost}/{queue_name}")
+    response.raise_for_status()
+    data = response.json()
+    return {metric: data[metric] for metric in METRICS}
+
+
+def main():
+    args = parse_args()
+    http = get_http_client(args.conf)
+    vhost = quote(args.vhost, safe="")
+    zabbix_sender_args = []
+    if args.tls_psk_identity:
+        zabbix_sender_args.extend(
+            [
+                "--tls-connect",
+                "psk",
+                "--tls-psk-identity",
+                args.tls_psk_identity,
+                "--tls-psk-file",
+                args.tls_psk_file,
+            ]
+        )
+    zabbix_sender_args.extend(["--host", args.host])
+    for queue_name in args.queue_name:
+        data = get_queue_values(http, vhost, queue_name)
+        for metric, value in data.items():
+            command = [
+                "zabbit_sender",
+                *zabbix_sender_args,
+                "--key",
+                f"rabbitmq.queue.{queue_name}.{metric}"
+                "--value",
+                str(value),
+            ]
+            print(" ".join(command))
+            # run(command, check=True)
+
+
+if __name__ == "__main__":
+    main()

roles/rabbitmq_cluster/files/send-rabbitmq-queue.service

@@ -0,0 +1,8 @@
+[Unit]
+Description=Send queue metrics to Zabbix
+
+[Service]
+Type=oneshot
+ExecStart=/usr/local/bin/send-rabbitmq-queue --tls-psk-identity foo --tls-psk-file bar centos-stream-robosignatory
+User=nrpe
+Group=nrpe

roles/rabbitmq_cluster/files/send-rabbitmq-queue.timer

@@ -0,0 +1,12 @@
+[Unit]
+Description=Send queue metrics to Zabbix
+ConditionPathExists=/etc/nrpe.d/rabbitmq_args.ini
+Wants=network-online.target
+
+[Timer]
+OnCalendar=5min
+Persistent=true
+
+[Install]
+WantedBy=timers.target
+

roles/rabbitmq_cluster/tasks/centos-monitoring.yml

@@ -0,0 +1,22 @@
+- name: Copy over the check script
+  ansible.builtin.copy:
+    src: send-rabbitmq-queue.py
+    dest: /usr/local/bin/send-rabbitmq-queue
+    mode: 0750
+    owner: root
+    group: nrpe
+
+- name: Setup systemd unit files
+  ansible.builtin.copy:
+    src: send-rabbitmq-queue.{{item}}
+    dest: /etc/systemd/system/send-rabbitmq-queue.{{item}}
+  with_items:
+    - service
+    - timer
+  notify:
+  - Reload systemd
+
+- name: Enable systemd timer
+  service:
+    name: send-rabbitmq-queue.timer
+    enabled: yes

roles/rabbitmq_cluster/tasks/main.yml

@@ -449,6 +449,13 @@
   ansible.builtin.command: /etc/nagios/selinux-load.sh
   when: selinux_module is changed
 
+# Queue monitoring for CentOS Zabbix
+- import_tasks: centos-monitoring.yml
+  when: inventory_hostname.startswith('rabbitmq03')
+  tags:
+  - rabbitmq_cluster
+  - config
+
 # Individual applications accounts & queues
 - import_tasks: apps.yml
   tags: