fix: alerting

hosts/amzebs-01/configuration.nix

@@ -60,6 +60,9 @@
     };
   };
 
+  # Systemd services to monitor
+  services.victoriametrics.monitoredServices = [ "mysql" "nginx" "phpfpm-.*" ];
+
   # backups - adjust repo for this host
   borgbackup.repo = "u149513-sub10@u149513-sub10.your-backup.de:borg";
 

hosts/fw/configuration.nix

@@ -76,6 +76,9 @@
 
   networkPrefix = "10.42";
 
+  # Systemd services to monitor
+  services.victoriametrics.monitoredServices = [ "ai-mailer" "container@git" "microvm@git-runner-" ];
+
   nixpkgs.overlays = [
     (import ./utils/overlays/packages.nix)
   ];

hosts/fw/modules/fwmetrics.nix

@@ -2,42 +2,19 @@
 let
   configure_prom = builtins.toFile "prometheus.yml" ''
     scrape_configs:
-    # System metrics
+    - job_name: 'server'
-    - job_name: 'node'
       stream_parse: true
       static_configs:
       - targets:
         - ${config.networking.hostName}:9100
-
-    # Systemd service monitoring
-    - job_name: 'systemd'
-      metrics_path: /metrics
-      params:
-        collect[]:
-          - 'systemd.service.state'
-          - 'systemd.service.start_time_seconds'
-          - 'systemd.unit_file.state'
-      static_configs:
-      - targets:
-        - ${config.networking.hostName}:9100
-      relabel_configs:
-        - source_labels: [__name__]
-          regex: 'node_systemd_unit_state'
-          action: keep
-        - source_labels: [name]
-          regex: '(ai-mailer|container@git|microvm@git-runner-).*\.service'
-          action: keep
   '';
 in {
   sops.secrets.victoria-agent-env = {
     sopsFile = ../utils/modules/victoriametrics/secrets.yaml;
   };
 
-  services.prometheus.exporters.node = {
+  services.prometheus.exporters.node.enable = true;
-    enable = true;
+
-    enabledCollectors = [ "systemd" ];
-  };
-  
   systemd.services.export-fw-to-prometheus = {
     path = with pkgs; [victoriametrics];
     enable = true;

hosts/mail/modules/metrics/default.nix

@@ -5,4 +5,7 @@
     ./postfix-exporter.nix
     ./dovecot-exporter.nix
   ];
+
+  # Systemd services to monitor
+  services.victoriametrics.monitoredServices = [ "postfix" "dovecot" "openldap" "wireguard-wg_cloonar" ];
 }

hosts/web-arm/modules/grafana/alerting/service/services_down.nix

@@ -9,10 +9,10 @@ let
     { name = "OpenLDAP";      service = "openldap.service";               instance = "mail:9100"; }
     { name = "Gitea";         service = "container@git.service";          instance = "fw:9100"; }
     { name = "Gitea Runner";  service = "microvm@git-runner-1.service";   instance = "fw:9100"; }
-    { name = "WireGuard";     service = "wireguard-wg_cloonar.service";   instance = "mail:9100"; }
+    { name = "WireGuard";     service = "wireguard-wg_cloonar.service";   instance = "fw:9100"; }
     { name = "MySQL";         service = "mysql.service";                  instance = "amzebs-01:9100"; }
     { name = "Nginx";         service = "nginx.service";                  instance = "amzebs-01:9100"; }
-    { name = "PHP-FPM";       service = "phpfpm-.*\\.service";            instance = "amzebs-01:9100"; }
+    { name = "PHP-FPM";       service = "phpfpm-.*[.]service";            instance = "amzebs-01:9100"; }
   ];
 
   # Extract host from instance (e.g., "fw:9100" -> "fw")
@@ -25,12 +25,17 @@ let
   isRegex = svc: lib.hasInfix ".*" svc || lib.hasInfix "\\" svc;
 
   # Build the PromQL expression
+  # For regex patterns: use min() to alert if ANY matching service is down
+  # For single services: use OR vector(0) to handle missing metrics
   mkExpr = svc:
     let
       nameMatch = if isRegex svc.service
         then "name=~\"${svc.service}\""
         else "name=\"${svc.service}\"";
-    in "node_systemd_unit_state{state=\"active\", ${nameMatch}, instance=\"${svc.instance}\"} OR on() vector(0)";
+      baseQuery = "node_systemd_unit_state{state=\"active\", ${nameMatch}, instance=\"${svc.instance}\"}";
+    in if isRegex svc.service
+      then "min(${baseQuery})"
+      else "${baseQuery} OR on() vector(0)";
 
   mkServiceAlert = svc: {
     uid = mkUid svc.name;

utils/modules/victoriametrics/default.nix

@@ -1,6 +1,9 @@
 { config, lib, pkgs, ... }:
 with lib;
 let
+  cfg = config.services.victoriametrics;
+  serviceRegex = concatStringsSep "|" cfg.monitoredServices;
+
   configure_prom = builtins.toFile "prometheus.yml" ''
     scrape_configs:
     # System metrics
@@ -27,13 +30,20 @@ let
           regex: 'node_systemd_unit_state'
           action: keep
         - source_labels: [name]
-          regex: '(container@git|microvm@git-runner-|postfix|dovecot|openldap|wireguard-wg_cloonar).*\.service'
+          regex: '(${serviceRegex}).*\.service'
           action: keep
 
-    ${concatStringsSep "\n" config.services.victoriametrics.extraScrapeConfigs}
+    ${concatStringsSep "\n" cfg.extraScrapeConfigs}
   '';
 in {
   options.services.victoriametrics = {
+    monitoredServices = mkOption {
+      type = types.listOf types.str;
+      default = [];
+      description = "List of systemd service name patterns to monitor (without .service suffix)";
+      example = [ "mysql" "nginx" "phpfpm-.*" ];
+    };
+
     extraScrapeConfigs = mkOption {
       type = types.listOf types.str;
       default = [];