alloy(role): pull in alloy role that works from fux

roles/alloy/defaults/main.yaml

@@ -1,29 +1,33 @@
+---
+alloy_cap_net_bind_service: false
+alloy_do_we_have_v6: "{{ (ansible_facts['all_ipv6_addresses'] | select('match', '^[23]') | list | length > 0) | ternary('true', 'false') }}"
+
 alloy_config_default: |
   prometheus.remote_write "default" {
     endpoint {
       url = "https://metrics.hamburg.ccc.de/api/v1/write"
       basic_auth {
         username = "chaos"
-        password = "chaos_password"
+        password = "{{ metrics__chaos_password }}"
       }
     }
   }
 
-  prometheus.relabel "common" {
+  prometheus.relabel "chaosknoten_common" {
     forward_to = [prometheus.remote_write.default.receiver]
     rule {
       target_label = "org"
-      replacement = "noorg"
+      replacement = "ccchh"
     }
     rule {
       target_label = "site"
-      replacement = "nosite"
+      replacement = "wieske"
     }
     rule {
       source_labels = ["instance"]
       target_label = "instance"
       regex  = "([^:]+)"
-      replacement = "${1}.hosts.test"
+      replacement = "${1}.hosts.hamburg.ccc.de"
       action = "replace"
     }
   }
@@ -38,7 +42,56 @@ alloy_config_default: |
 
   prometheus.scrape "scrape_metrics" {
     targets           = prometheus.exporter.unix.local_system.targets
-    forward_to        = [prometheus.relabel.common.receiver]
+    forward_to        = [prometheus.relabel.chaosknoten_common.receiver]
   }
 
 alloy_config_additional: ""
+
+alloy_blackbox_exporter_config_default: |
+  modules:
+    https4_2xx:
+      prober: http
+      http:
+        valid_status_codes: []  # Defaults to 2xx
+        method: GET
+        follow_redirects: true
+        fail_if_ssl: false
+        fail_if_not_ssl: true
+        preferred_ip_protocol: "ip4"
+        ip_protocol_fallback: false
+
+    https6_2xx:
+      prober: http
+      http:
+        valid_status_codes: []  # Defaults to 2xx
+        method: GET
+        follow_redirects: true
+        fail_if_ssl: false
+        fail_if_not_ssl: true
+        preferred_ip_protocol: "ip6"
+        ip_protocol_fallback: false
+
+    tcp4_connect:
+      prober: tcp
+      tcp:
+        preferred_ip_protocol: "ip4"
+        ip_protocol_fallback: false
+
+    icmp4:
+      prober: icmp
+      icmp:
+        preferred_ip_protocol: "ip4"
+        ip_protocol_fallback: false
+
+    icmp6:
+      prober: icmp
+      icmp:
+        preferred_ip_protocol: "ip6"
+        ip_protocol_fallback: false
+
+alloy_blackbox_exporter_config_additional: ""
+
+alloy_unpoller_enabled: false
+alloy_unpoller_unifi_user: unpoller
+alloy_unpoller_unifi_password:
+alloy_unpoller_unifi_url:

roles/alloy/handlers/main.yml

@@ -0,0 +1,31 @@
+---
+- name: Alloy.enabled
+  become: true
+  ansible.builtin.systemd:
+    name: alloy
+    enabled: true
+
+- name: Alloy.restarted
+  become: true
+  ansible.builtin.systemd:
+    name: alloy
+    state: restarted
+    enabled: true
+
+- name: Alloy.reloaded
+  become: true
+  ansible.builtin.systemd:
+    name: alloy
+    state: reloaded
+    enabled: true
+
+- name: Systemd.daemon_reload
+  become: true
+  ansible.builtin.systemd_service:
+    daemon_reload: true
+
+- name: Unpoller.restarted
+  become: true
+  ansible.builtin.systemd_service:
+    name: unpoller.service
+    state: restarted

roles/alloy/tasks/blackbox-exporter.yaml

@@ -0,0 +1,11 @@
+---
+- name: Render blackbox exporter config
+  tags: [ blackbox-exporter, alloy ]
+  become: true
+  ansible.builtin.template:
+    src: config.blackbox-exporter.j2
+    dest: /etc/alloy/blackbox-exporter.yaml
+    mode: "0644"
+    owner: root
+  notify:
+    - Alloy.reloaded

roles/alloy/tasks/install_debian.yml

@@ -0,0 +1,20 @@
+---
+- name: Add grafana apt repo with signing key
+  become: true
+  ansible.builtin.deb822_repository:
+    name: grafana
+    types: deb
+    uris: https://apt.grafana.com
+    suites: stable
+    components:
+      - main
+    signed_by: https://apt.grafana.com/gpg.key
+
+- name: Run cache update
+  ansible.builtin.apt:
+    update_cache: true
+
+- name: Install alloy
+  become: true
+  ansible.builtin.package:
+    name: alloy

roles/alloy/tasks/main.yaml

@@ -1,50 +1,79 @@
-# https://github.com/grafana/grafana-ansible-collection/blob/main/roles/alloy/tasks/deploy.yml#L124
+---
-- name: ensure alloy user exists
+- name: Deploy Grafana Alloy with config
-  ansible.builtin.user:
+  tags: [ alloy ]
-    name: alloy
-    system: true
-    append: true
-    create_home: false
-    state: present
-
-- name: ensure the `/etc/alloy/` config directory exists
-  ansible.builtin.file:
-    path: /etc/alloy
-    state: directory
-    mode: "0770"
-    owner: root
-    group: alloy
-  become: true
-
-- name: synchronize the additional configuration files directory, if present
-  when: alloy__additional_configs_dir is defined and alloy__additional_configs_dir != ""
   block:
-    - name: ensure rsync is installed
+    - name: Install Alloy on Debian
-      ansible.builtin.apt:
+      when: ansible_facts['distribution'] == "Debian"
-        name: rsync
+      ansible.builtin.include_tasks: install_debian.yml
-      become: true
 
-    - name: synchronize the additional configuration files directory, if present
+    - name: Create systemd override for alloy
-      ansible.posix.synchronize:
-        src: "{{ alloy__additional_configs_dir }}"
-        dest: /etc/alloy/additional
-        delete: true
-        recursive: true
-        use_ssh_args: true
-        rsync_opts:
-          - "--chown=root:alloy"
       become: true
-
+      when: alloy_cap_net_bind_service
-- name: delete the additional configuration files directory, if not present
+      notify: Systemd.daemon_reload
-  when: alloy__additional_configs_dir is not defined or alloy__additional_configs_dir == ""
+      block:
+        - name: Create systemd override directory
           ansible.builtin.file:
-    path: /etc/alloy/additional
+            path: /etc/systemd/system/alloy.service.d
-    state: absent
+            state: directory
-  become: true
+            mode: "0755"
+            owner: root
+            group: root
+        - name: Add CAP_NET_BIND_SERVICE capability
+          ansible.builtin.copy:
+            dest: /etc/systemd/system/alloy.service.d/override.conf
+            mode: "0755"
+            owner: root
+            group: root
+            content: |
+              [Service]
+              AmbientCapabilities=CAP_NET_BIND_SERVICE
 
-- name: Setup Alloy
+    - name: Deploy the Alloy config file
-  ansible.builtin.import_role:
-    name: grafana.grafana.alloy
-  vars:
-    alloy_config: "{{ alloy_config_default ~ alloy_config_additional }}"
       become: true
+      ansible.builtin.template:
+        src: "config.alloy.j2"
+        dest: "/etc/alloy/config.alloy"
+        owner: "alloy"
+        group: "alloy"
+        mode: "0640"
+      notify:
+        - Alloy.restarted
+        - Alloy.enabled
+
+    - name: Validate Alloy configuration and print error if failed
+      block:
+        - name: Validate Alloy configuration
+          become: true
+          register: alloy_validation
+          changed_when: false
+          ansible.builtin.command: alloy validate /etc/alloy/config.alloy
+      rescue:
+        - name: Show error if validation failed
+          when: alloy_validation.rc != 0
+          ansible.builtin.fail:
+            msg: |
+              Alloy configuration validation failed!
+              Return code: {{ alloy_validation.rc }}
+              Stdout:
+              {{ alloy_validation.stdout }}
+              Stderr:
+              {{ alloy_validation.stderr }}
+
+    - name: Deploy blackbox-exporter config
+      when: alloy_enable_blackbox_exporter
+      ansible.builtin.include_tasks: blackbox-exporter.yaml
+    - name: Deploy unpoller exporter
+      when: alloy_unpoller_enabled
+      ansible.builtin.include_tasks: unpoller.yaml
+
+    - name: Flush handlers
+      ansible.builtin.meta: flush_handlers
+
+    - name: Ensure that Alloy is working
+      ansible.builtin.uri:
+        url: "http://localhost:12345/-/ready"
+        method: GET
+      register: alloy_status_code
+      retries: 8
+      delay: 6
+      until: not alloy_status_code.failed

roles/alloy/tasks/unpoller.yaml

@@ -0,0 +1,76 @@
+---
+- name: Install and Setup unpoller
+  tags: [ unpoller, alloy, podman, container ]
+  block:
+    - name: Install Podman
+      become: true
+      ansible.builtin.package:
+        name:
+          - podman
+          - fuse-overlayfs
+
+    - name: Create unpoller group
+      become: true
+      register: alloy_unpoller_group
+      ansible.builtin.group:
+        name: unpoller
+
+    - name: Create unpoller user
+      become: true
+      register: alloy_unpoller_user
+      ansible.builtin.user:
+        name: unpoller
+        group: unpoller
+        system: true
+
+    - name: Create unpoller configuration directory
+      become: true
+      ansible.builtin.file:
+        path: /etc/unpoller
+        state: directory
+        owner: "{{ alloy_unpoller_user.name }}"
+        mode: "0755"
+
+    - name: Render unpoller configuration
+      become: true
+      notify: Unpoller.restarted
+      ansible.builtin.template:
+        src: unpoller.conf.j2
+        dest: /etc/unpoller/up.conf
+        owner: "{{ alloy_unpoller_user.name }}"
+        mode: "0644"
+
+    - name: Create /etc/containers/systemd directory
+      become: true
+      ansible.builtin.file:
+        path: /etc/containers/systemd
+        state: directory
+        owner: root
+        mode: "0755"
+
+    - name: Configure unpoller quadlet
+      become: true
+      notify: Unpoller.restarted
+      register: alloy_unpoller_quadlet
+      vars:
+        container_image_with_tag: ghcr.io/unpoller/unpoller:v2
+        container_unit_args:
+        container_exec:
+        container_envs: [ ]
+        container_volumes:
+          - { host: "/etc/unpoller", inside: "/etc/unpoller:ro" }
+        user_id: "{{ alloy_unpoller_user.uid }}"
+        group_id: "{{ alloy_unpoller_group.gid }}"
+      ansible.builtin.template:
+        src: podman.container.jinja
+        dest: /etc/containers/systemd/unpoller.container
+        owner: root
+        mode: "0644"
+
+    - name: Ensure unpoller is running and enabled
+      become: true
+      ansible.builtin.systemd:
+        name: unpoller.service
+        state: started
+        enabled: true
+        daemon_reload: "{{ alloy_unpoller_quadlet.changed }}"

roles/alloy/templates/config.alloy.j2

@@ -0,0 +1,3 @@
+// Ansible Managed
+{{ alloy_config_default }}
+{{ alloy_config_additional }}

roles/alloy/templates/config.blackbox-exporter.j2

@@ -0,0 +1,3 @@
+# {{  ansible_managed }}
+{{ alloy_blackbox_exporter_config_default }}
+{{ alloy_blackbox_exporter_config_additional }}

roles/alloy/templates/podman.container.jinja

@@ -0,0 +1,29 @@
+# {{ ansible_managed }}
+# man podman-systemd.unit
+
+{% if container_unit_args is not none %}
+[Unit]
+{% for container_unit_arg in container_unit_args %}
+{{ container_unit_arg }}
+{% endfor %}
+{% endif %}
+
+[Container]
+Image={{ container_image_with_tag }}
+{% if container_exec is not none %}
+Exec={{ container_exec }}
+{% endif %}
+{% for container_env in container_envs %}
+Environment={{ container_env.key }}={{ container_env.value }}
+{% endfor %}
+Network=host
+UserNS=host
+User={{ user_id }}
+Group={{ group_id }}
+{% for container_vol in container_volumes %}
+Volume={{ container_vol.host }}:{{ container_vol.inside }}
+{% endfor %}
+LogDriver=journald
+
+[Install]
+WantedBy=default.target

roles/alloy/templates/unpoller.conf.j2

@@ -0,0 +1,248 @@
+# {{ ansible_managed }}
+# Unpoller v2 primary configuration file. TOML FORMAT #
+###########################################################
+
+[poller]
+  # Turns on line numbers, microsecond logging, and a per-device log.
+  # The default is false, but I personally leave this on at home (four devices).
+  # This may be noisy if you have a lot of devices. It adds one line per device.
+  debug = false
+
+  # Turns off per-interval logs. Only startup and error logs will be emitted.
+  # Recommend enabling debug with this setting for better error logging.
+  quiet = false
+
+  # Load dynamic plugins. Advanced use; only sample mysql plugin provided by default.
+  plugins = []
+
+#### OUTPUTS
+
+    # If you don't use an output, you can disable it.
+
+[prometheus]
+  disable = false
+  # This controls on which ip and port /metrics is exported when mode is "prometheus".
+  # This has no effect in other modes. Must contain a colon and port.
+  http_listen = "0.0.0.0:9130"
+  # Adding an SSL Cert and Cert Key will make Poller listen with SSL/https.
+  ssl_cert_path = ""
+  ssl_key_path  = ""
+  # Errors are rare. Setting this to true will report them to Prometheus.
+  report_errors = false
+  ## Record data for disabled or down (unlinked) switch ports.
+  dead_ports = false
+
+[influxdb]
+  disable = true
+  # InfluxDB does not require auth by default, so the user/password are probably unimportant.
+  url  = "http://127.0.0.1:8086"
+  user = "unifipoller"
+  # Password for InfluxDB user (above).
+  # If the password provided here begins with file:// then the password is read in from
+  # the file path that follows the file:// prefix. ex: file:///etc/influxdb/passwd.file
+  pass = "unifipoller"
+  # Be sure to create this database. See the InfluxDB Wiki page for more info.
+  db = "unifi"
+  # If your InfluxDB uses a valid SSL cert, set this to true.
+  verify_ssl = false
+  # The UniFi Controller only updates traffic stats about every 30 seconds.
+  # Setting this to something lower may lead to "zeros" in your data.
+  # If you're getting zeros now, set this to "1m"
+  interval = "30s"
+  ## Record data for disabled or down (unlinked) switch ports.
+  dead_ports = false
+
+# To enable output of UniFi Events to Loki, add a URL; it's disabled otherwise.
+# User, pass and tenant_id are optional and most folks wont set them.
+# Pick which logs you want per-controller in the [unifi.controller] section.
+# This is a new feature. Feedback welcome!
+[loki]
+  disable = true
+  url = ""
+  # The rest of this is advanced & optional. See wiki.
+  user       = ""
+  pass       = ""
+  verify_ssl = false
+  tenant_id  = ""
+  interval   = "2m"
+  timeout    = "10s"
+
+[datadog]
+  # How often to poll UniFi and report to Datadog.
+  interval = "2m"
+
+  # To enable this output plugin
+  enable = false
+
+  # Datadog Custom Options
+
+  # address to talk to the datadog agent, by default this uses the local statsd UDP interface
+  # address = "localhost:8125"
+
+  # namespace to prepend to all data, default is no additional prefix.
+  # namespace = ""
+
+  # tags to append to all data
+  # tags = [ "customer:abc_corp" ]
+
+  # For more advanced options for very large amount of data collected see the upstream
+  # github.com/unpoller/unpoller/pkg/datadogunifi repository README.
+
+
+# Unpoller has an optional web server. To turn it on, set enable to true. If you
+# wish to use SSL, provide SSL cert and key paths. This interface is currently
+# read-only; it just displays information, like logs, devices and clients.
+# Notice: Enabling the web server with many sites will increase memory usage.
+# This is a new feature and lacks a UI, enabling only recommended for testing.
+[webserver]
+  enable = false
+  port   = 37288
+  # The HTML path is different on Windows and BSD/macOS.
+  html_path     = "/usr/local/lib/unpoller/web"
+  ssl_cert_path = ""
+  ssl_key_path  = ""
+  # How many events per event group to hold. 200-2000. Use fewer with many sites.
+  # With 1 site, you'll have a max total of 9 event groups; 1 per plugin, 4 per site.
+  # Each site adds 1 event group for each of these inputs that is enabled:
+  #   save_ids, save_events, save_anomalies, save_alarms.
+  max_events = 200
+
+# By default the web interface does not require authentication. You can change
+# that by adding a username and password hash (or multiple) below.
+# To create a hash, run unifi-poller with the -e CLI argument. See Wiki for more!
+[webserver.accounts]
+# username = "password-hash"
+# captain = "$2a$04$mxw6i0LKH6u46oaLK2cq5eCTAAFkfNiRpzNbz.EyvJZZWNa2FzIlS"
+
+#### INPUTS
+
+[unifi]
+  # Setting this to true and providing default credentials allows you to skip
+  # configuring controllers in this config file. Instead you configure them in
+  # your prometheus.yml config. Prometheus then sends the controller URL to
+  # Unpoller when it performs the scrape. This is useful if you have many,
+  # or changing controllers. See wiki for more.
+  dynamic = false
+
+# The following section contains the default credentials/configuration for any
+# dynamic controller (see above section), or the primary controller if you do not
+# provide one and dynamic is disabled. In other words, you can just add your
+# controller here and delete the following section. The internal defaults are
+# shown below. Any missing values will assume these displayed defaults.
+
+[unifi.defaults]
+  # URL for the UniFi Controller. Do not add any paths after the host:port.
+  # Do not use port 8443 if you have a UDM; just use "https://ip".
+  url = "https://invalid.invalid:8443"
+
+  # Make a read-only user in the UniFi Admin Settings, allow it access to all sites.
+  user = "{{ alloy_unpoller_unifi_user }}"
+
+  # Password for UniFi controller user (above).
+  # If the password provided here begins with file:// then the password is read in from
+  # the file path that follows the file:// prefix. ex: file:///etc/unifi/password.file
+  # ex: file:///etc/unifi/passwd.file, windows: file://C:\\UserData\\Unifi\\Passwd.txt
+  pass = ""
+
+  # API Key
+  # Unifi now supports API Key authentication. This is exclusive of user/pass auth.
+  # api_key = "unifiapikey"
+
+  # If the controller has more than one site, specify which sites to poll here.
+  # Set this to ["default"] to poll only the first site on the controller.
+  # A setting of ["all"] will poll all sites; this works if you only have 1 site too.
+  sites = ["all"]
+
+  # Added an example for overriding the default site name.
+  # default_site_name_override = "My Custom Default Site"
+
+  # Specify a timeout, leave missing to declare infinite wait. This determines the maximum
+  # time to wait for a response from the unifi controller on any API request.
+  # timeout = 60s
+
+  # Enable collection of site data. This data powers the Network Sites dashboard.
+  # It's not valuable to everyone and setting this to false will save resources.
+  save_sites = false
+
+  # Hash, with md5, client names and MAC addresses. This attempts to protect
+  # personally identifiable information. Most users won't want to enable this.
+  hash_pii = false
+
+  # Enable collection of Intrusion Detection System Data (InfluxDB/Loki only).
+  # Only useful if IDS or IPS are enabled on one of the sites. This may store
+  # a lot of information. Only recommended for testing and debugging. There
+  # may not be any dashboards to display this data. It can be used for annotations.
+  # Enable this only if using InfluxDB or Loki. This will leak PII data!
+  save_ids = false
+
+  # Enable collection of UniFi Events (InfluxDB/Loki only).
+  # This may store a lot of information. Only recommended for testing and debugging.
+  # There are no dashboards to display this data. It can be used for annotations.
+  # This is a new (June, 2020) feature. Please provide feedback if you try it out!
+  # Enable this only if using InfluxDB or Loki. This will leak PII data!
+  save_events = false
+
+  # Enable collection of UniFi Alarms (InfluxDB/Loki only).
+  # There are no dashboards to display this data. It can be used for annotations.
+  # This is a new (June, 2020) feature. Please provide feedback if you try it out!
+  # Enable this only if using InfluxDB or Loki. This will leak PII data!
+  save_alarms = false
+
+  # Enable collection of UniFi Anomalies (InfluxDB/Loki only).
+  # There are no dashboards to display this data. It can be used for annotations.
+  # This is a new (June, 2020) feature. Please provide feedback if you try it out!
+  # Enable this only if using InfluxDB or Loki.
+  save_anomalies = false
+
+  # Enable collection of Deep Packet Inspection data. This data breaks down traffic
+  # types for each client and site, it powers a dedicated DPI dashboard.
+  # Enabling this adds roughly 150 data points per client.  That's 6000 metrics for
+  # 40 clients.  This adds a little bit of poller run time per interval and causes
+  # more API requests to your controller(s). Don't let these "cons" sway you:
+  # it's cool data. Please provide feedback on your experience with this feature.
+  save_dpi = false
+
+  ## Enabling save_rogue stores even more data in your time series databases.
+  ## This saves neighboring access point metrics in a dedicated table or namespace.
+  save_rogue = false
+
+  # If your UniFi controller has a valid SSL certificate (like lets encrypt),
+  # you can enable this option to validate it. Otherwise, any SSL certificate is
+  # valid. If you don't know if you have a valid SSL cert, then you don't have one.
+  verify_ssl = false
+
+  ## You may provide a list of SSL cert files (PEM format) that you expect your
+  ## controller to use. As long as one of the certs you provide here shows up in
+  ## the cert trust chain the controller presents it will be accepted and allowed.
+  ## These files may be re-read while poller is running.
+  ## Example: ssl_cert_paths = ["/path/to/cert.pem", "/another/cert.pem"]
+  ssl_cert_paths = []
+
+# The following is optional and used for configurations with multiple UniFi controllers.
+
+# You may repeat the following [[unifi.controller]] section as many times as needed to
+# poll multiple controllers. Uncomment the entire section including [[unifi.controller]].
+# Omitted variables will have their values taken from the defaults, above.
+#
+#[[unifi.controller]]
+#  url         = "https://127.0.0.1:8443"
+#  user        = "unifipoller"
+#  pass        = "unifipoller"
+#  sites       = ["all"]
+#  save_sites  = true
+#  hash_pii    = false
+#  save_ids    = false
+#  save_events = false
+#  save_alarms = false
+#  save_anomalies = false
+#  save_dpi    = false
+#  save_rogue  = false
+#  verify_ssl  = false
+#  ssl_cert_paths = []
+
+[[unifi.controller]]
+url = "{{ alloy_unpoller_unifi_url | mandatory }}"
+user = "{{ alloy_unpoller_unifi_user }}"
+pass = "{{ alloy_unpoller_unifi_password | mandatory }}"
+hash_pii = true
+save_anomalies = false

roles/alloy/vars/main.yaml

@@ -0,0 +1,2 @@
+---
+alloy_enable_blackbox_exporter: "{{ 'prometheus.exporter.blackbox' in (alloy_config_default + alloy_config_additional) }}"