mio-ops/profiles/prometheus.nix

# NixOps configuration for the hosts running Prometheus on a Cardano node

{ config, pkgs, lib, ... }:

{

  services = {
    prometheus = {
      enable = true;
      webExternalUrl = "https://monitoring.mcwhirter.io/prometheus/";
      extraFlags = [
        "--storage.tsdb.retention.time 8760h"
      ];
      exporters = {
        node = {
          enable = true;
          openFirewall = true;
          enabledCollectors = [
            "systemd"
            "tcpstat"
            "conntrack"
            "diskstats"
            "entropy"
            "filefd"
            "filesystem"
            "loadavg"
            "meminfo"
            "netdev"
            "netstat"
            "stat"
            "time"
            "ntp"
            "timex"
            "vmstat"
            "logind"
            "interrupts"
            "ksmd"
            "processes"
          ];
        };
      };
      #alertmanager = {
      #  enable = true;
      #  webExternalUrl = "https://monitoring.mcwhirter.io/alertmanager/";
      #  configuration = [
      #  ];
      #};
      #alertmanagers = [ {
      #  scheme = "http";
      #  path_prefix = "/";
      #  static_configs = [ {
      #    targets = [ "airgead.mcwhirter.io:9093" ];
      #  } ];
      #} ];
      rules = [ (builtins.toJSON {
        groups = [
          {
            name = "system";
            rules = [
              {
                alert = "node_down";
                expr = "up == 0";
                for = "5m";
                labels = {
                  severity = "page";
                };
                annotations = {
                  summary = "{{$labels.alias}}: Node is down.";
                  description = "{{$labels.alias}} has been down for more than 5 minutes.";
                };
              }
              {
                alert = "node_systemd_service_failed";
                expr = "node_systemd_unit_state{state=\"failed\"} == 1";
                for = "4m";
                labels = {
                  severity = "page";
                };
                annotations = {
                  summary = "{{$labels.alias}}: Service {{$labels.name}} failed to start.";
                  description = "{{$labels.alias}} failed to (re)start service {{$labels.name}}.";
                };
              }
              {
                alert = "node_filesystem_full_90percent";
                expr = "sort(node_filesystem_free_bytes{device!=\"ramfs\"} < node_filesystem_size_bytes{device!=\"ramfs\"} * 0.1) / 1024^3";
                for = "5m";
                labels = {
                  severity = "page";
                };
                annotations = {
                  summary = "{{$labels.alias}}: Filesystem is running out of space soon.";
                  description = "{{$labels.alias}} device {{$labels.device}} on {{$labels.mountpoint}} got less than 10% space left on its filesystem.";
                };
              }
              {
                alert = "node_filesystem_full_in_4h";
                expr = "predict_linear(node_filesystem_free_bytes{device!=\"ramfs\",device!=\"tmpfs\",fstype!=\"autofs\",fstype!=\"cd9660\"}[4h], 4*3600) <= 0";
                for = "5m";
                labels = {
                  severity = "page";
                };
                annotations = {
                  summary = "{{$labels.alias}}: Filesystem is running out of space in 4 hours.";
                  description = "{{$labels.alias}} device {{$labels.device}} on {{$labels.mountpoint}} is running out of space of in approx. 4 hours";
                };
              }
              {
                alert = "node_filedescriptors_full_in_3h";
                expr = "predict_linear(node_filefd_allocated[1h], 3*3600) >= node_filefd_maximum";
                for = "20m";
                labels = {
                  severity = "page";
                };
                annotations = {
                  summary = "{{$labels.alias}} is running out of available file descriptors in 3 hours.";
                  description = "{{$labels.alias}} is running out of available file descriptors in approx. 3 hours";
                };
              }
              {
                alert = "node_load1_90percent";
                expr = "node_load1 / on(alias) count(node_cpu_seconds_total{mode=\"system\"}) by (alias) >= 0.9";
                for = "1h";
                labels = {
                  severity = "page";
                };
                annotations = {
                  summary = "{{$labels.alias}}: Running on high load.";
                  description = "{{$labels.alias}} is running with > 90% total load for at least 1h.";
                };
              }
              {
                alert = "node_cpu_util_90percent";
                expr = "100 - (avg by (alias) (irate(node_cpu_seconds_total{mode=\"idle\"}[5m])) * 100) >= 90";
                for = "1h";
                labels = {
                  severity = "page";
                };
                annotations = {
                  summary = "{{$labels.alias}}: High CPU utilization.";
                  description = "{{$labels.alias}} has total CPU utilization over 90% for at least 1h.";
                };
              }
              {
                alert = "node_ram_using_99percent";
                expr = "node_memory_MemFree_bytes + node_memory_Buffers_bytes + node_memory_Cached_bytes < node_memory_MemTotal_bytes * 0.01";
                for = "30m";
                labels = {
                  severity = "page";
                };
                annotations = {
                  summary = "{{$labels.alias}}: Using lots of RAM.";
                  description = "{{$labels.alias}} is using at least 90% of its RAM for at least 30 minutes now.";
                };
              }
              {
                alert = "node_swap_using_80percent";
                expr = "node_memory_SwapTotal_bytes - (node_memory_SwapFree_bytes + node_memory_SwapCached_bytes) > node_memory_SwapTotal_bytes * 0.8";
                for = "10m";
                labels = {
                  severity = "page";
                };
                annotations = {
                  summary = "{{$labels.alias}}: Running out of swap soon.";
                  description = "{{$labels.alias}} is using 80% of its swap space for at least 10 minutes now.";
                };
              }
              {
                alert = "node_time_unsync";
                expr = "abs(node_timex_offset_seconds) > 0.050 or node_timex_sync_status != 1";
                for = "1m";
                labels = {
                  severity = "page";
                };
                annotations = {
                  summary = "{{$labels.alias}}: Clock out of sync with NTP";
                  description = "{{$labels.alias}} Local clock offset is too large or out of sync with NTP";
                };
              }
            ];
          }
        ];
      })];
      scrapeConfigs = [
        {
          job_name = "prometheus";
          scrape_interval = "5s";
          static_configs = [
            {
              targets = [
                "localhost:9090"
              ];
              labels = { alias = "prometheus"; };
            }
          ];
        }
        {
          job_name = "cardano-node";
          scrape_interval = "10s";
          static_configs = [
            {
              targets = [ "127.0.0.1:12798" ];
              labels = { alias = "airgead"; };
            }
          ];
        }
        {
          job_name = "node";
          scrape_interval = "10s";
          static_configs = [
            {
              targets = [
                "airgead.mcwhirter.io:9100"
              ];
              labels = {
                alias = "airgead.mcwhirter.io";
              };
            }
          ];
        }
      ];
    };

  };

}