From 9cbce7b82f0bad8efb4a2d47e5921f93ea9dc004 Mon Sep 17 00:00:00 2001 From: lassulus Date: Sat, 8 Jul 2023 16:20:51 +0200 Subject: l: add monitoring config --- lass/2configs/monitoring/alert-rules.nix | 208 +++++++++++++++++++++++++++++++ lass/2configs/monitoring/prometheus.nix | 110 ++++++++++++++++ lass/2configs/monitoring/telegraf.nix | 72 +++++++++++ 3 files changed, 390 insertions(+) create mode 100644 lass/2configs/monitoring/alert-rules.nix create mode 100644 lass/2configs/monitoring/prometheus.nix create mode 100644 lass/2configs/monitoring/telegraf.nix (limited to 'lass') diff --git a/lass/2configs/monitoring/alert-rules.nix b/lass/2configs/monitoring/alert-rules.nix new file mode 100644 index 00000000..eae2569f --- /dev/null +++ b/lass/2configs/monitoring/alert-rules.nix @@ -0,0 +1,208 @@ +# inspiration from https://github.com/Mic92/dotfiles/blob/master/nixos/eva/modules/prometheus/alert-rules.nix +{ lib }: + +lib.mapAttrsToList + (name: opts: { + alert = name; + expr = opts.condition; + for = opts.time or "2m"; + labels = { }; + annotations.description = opts.description; + }) + ({ + prometheus_too_many_restarts = { + condition = ''changes(process_start_time_seconds{job=~"prometheus|pushgateway|alertmanager|telegraf"}[15m]) > 2''; + description = "Prometheus has restarted more than twice in the last 15 minutes. It might be crashlooping."; + }; + + alert_manager_config_not_synced = { + condition = ''count(count_values("config_hash", alertmanager_config_hash)) > 1''; + description = "Configurations of AlertManager cluster instances are out of sync."; + }; + + prometheus_not_connected_to_alertmanager = { + condition = "prometheus_notifications_alertmanagers_discovered < 1"; + description = "Prometheus cannot connect the alertmanager\n VALUE = {{ $value }}\n LABELS = {{ $labels }}"; + }; + + prometheus_rule_evaluation_failures = { + condition = "increase(prometheus_rule_evaluation_failures_total[3m]) > 0"; + description = "Prometheus encountered {{ $value }} rule evaluation failures, leading to potentially ignored alerts.\n VALUE = {{ $value }}\n LABELS = {{ $labels }}"; + }; + + prometheus_template_expansion_failures = { + condition = "increase(prometheus_template_text_expansion_failures_total[3m]) > 0"; + time = "0m"; + description = "Prometheus encountered {{ $value }} template text expansion failures\n VALUE = {{ $value }}\n LABELS = {{ $labels }}"; + }; + + promtail_request_errors = { + condition = ''100 * sum(rate(promtail_request_duration_seconds_count{status_code=~"5..|failed"}[1m])) by (namespace, job, route, instance) / sum(rate(promtail_request_duration_seconds_count[1m])) by (namespace, job, route, instance) > 10''; + time = "15m"; + description = ''{{ $labels.job }} {{ $labels.route }} is experiencing {{ printf "%.2f" $value }}% errors.''; + }; + + promtail_file_lagging = { + condition = ''abs(promtail_file_bytes_total - promtail_read_bytes_total) > 1e6''; + time = "15m"; + description = ''{{ $labels.instance }} {{ $labels.job }} {{ $labels.path }} has been lagging by more than 1MB for more than 15m.''; + }; + + filesystem_full_80percent = { + condition = ''disk_used_percent{mode!="ro"} >= 95''; + time = "10m"; + description = "{{$labels.instance}} device {{$labels.device}} on {{$labels.path}} got less than 20% space left on its filesystem."; + }; + + filesystem_full_krebs = { + condition = ''disk_used_percent{mode!="ro", org="krebs"} >= 95''; + time = "10m"; + description = "{{$labels.instance}} device {{$labels.device}} on {{$labels.path}} got less than 5% space left on its filesystem."; + }; + + filesystem_inodes_full = { + condition = ''disk_inodes_free / disk_inodes_total < 0.10''; + time = "10m"; + description = "{{$labels.instance}} device {{$labels.device}} on {{$labels.path}} got less than 10% inodes left on its filesystem."; + }; + + daily_task_not_run = { + # give 6 hours grace period + condition = ''time() - task_last_run{state="ok",frequency="daily"} > (24 + 6) * 60 * 60''; + description = "{{$labels.host}}: {{$labels.name}} was not run in the last 24h"; + }; + + daily_task_failed = { + condition = ''task_last_run{state="fail"}''; + description = "{{$labels.host}}: {{$labels.name}} failed to run"; + }; + + swap_using_30percent = { + condition = "mem_swap_total - (mem_swap_cached + mem_swap_free) > mem_swap_total * 0.3"; + time = "30m"; + description = "{{$labels.host}} is using 30% of its swap space for at least 30 minutes."; + }; + + systemd_service_failed = { + condition = ''systemd_units_active_code{name!~"nixpkgs-update-.*.service"} == 3''; + description = "{{$labels.host}} failed to (re)start service {{$labels.name}}."; + }; + + service_not_running = { + condition = ''systemd_units_active_code{name=~"teamspeak3-server.service|tt-rss.service", sub!="running"}''; + description = "{{$labels.host}} should have a running {{$labels.name}}."; + }; + + nfs_export_not_present = { + condition = "nfs_export_present == 0"; + time = "1h"; + description = "{{$labels.host}} cannot reach nfs export [{{$labels.server}}]:{{$labels.path}}"; + }; + + ram_using_90percent = { + condition = "mem_buffered + mem_free + mem_cached < mem_total * 0.1"; + time = "1h"; + description = "{{$labels.host}} is using at least 90% of its RAM for at least 1 hour."; + }; + load15 = { + condition = ''system_load15 / system_n_cpus{org!="nix-community"} >= 2.0''; + time = "10m"; + description = "{{$labels.host}} is running with load15 > 1 for at least 5 minutes: {{$value}}"; + }; + reboot = { + condition = "system_uptime < 300"; + description = "{{$labels.host}} just rebooted."; + }; + uptime = { + # too scared to upgrade matchbox + condition = ''system_uptime {host!~"^(matchbox|grandalf)$"} > 2592000''; + description = "Uptime monster: {{$labels.host}} has been up for more than 30 days."; + }; + telegraf_down = { + condition = ''min(up{job=~"telegraf",type!='mobile'}) by (source, job, instance, org) == 0''; + time = "3m"; + description = "{{$labels.instance}}: {{$labels.job}} telegraf exporter from {{$labels.source}} is down."; + }; + ping = { + condition = "ping_result_code{type!='mobile'} != 0"; + description = "{{$labels.url}}: ping from {{$labels.instance}} has failed!"; + }; + ping_high_latency = { + condition = "ping_average_response_ms{type!='mobile'} > 5000"; + description = "{{$labels.instance}}: ping probe from {{$labels.source}} is encountering high latency!"; + }; + http = { + condition = "http_response_result_code != 0"; + description = "{{$labels.server}} : http request failed from {{$labels.instance}}: {{$labels.result}}!"; + }; + http_match_failed = { + condition = "http_response_response_string_match == 0"; + description = "{{$labels.server}} : http body not as expected; status code: {{$labels.status_code}}!"; + }; + dns_query = { + condition = "dns_query_result_code != 0"; + description = "{{$labels.domain}} : could retrieve A record {{$labels.instance}} from server {{$labels.server}}: {{$labels.result}}!"; + }; + secure_dns_query = { + condition = "secure_dns_state != 0"; + description = "{{$labels.domain}} : could retrieve A record {{$labels.instance}} from server {{$labels.server}}: {{$labels.result}} for protocol {{$labels.protocol}}!"; + }; + connection_failed = { + condition = "net_response_result_code != 0"; + description = "{{$labels.server}}: connection to {{$labels.port}}({{$labels.protocol}}) failed from {{$labels.instance}}"; + }; + healthchecks = { + condition = "hc_check_up == 0"; + description = "{{$labels.instance}}: healtcheck {{$labels.job}} fails!"; + }; + cert_expiry = { + condition = "x509_cert_expiry < 7*24*3600"; + description = "{{$labels.instance}}: The TLS certificate from {{$labels.source}} will expire in less than 7 days: {{$value}}s"; + }; + + postfix_queue_length = { + condition = "avg_over_time(postfix_queue_length[1h]) > 10"; + description = "{{$labels.instance}}: postfix mail queue has undelivered {{$value}} items"; + }; + + zfs_errors = { + condition = "zfs_arcstats_l2_io_error + zfs_dmu_tx_error + zfs_arcstats_l2_writes_error > 0"; + description = "{{$labels.instance}} reports: {{$value}} ZFS IO errors."; + }; + + # ignore devices that disabled S.M.A.R.T (example if attached via USB) + smart_errors = { + condition = ''smart_device_health_ok{enabled!="Disabled"} != 1''; + description = "{{$labels.instance}}: S.M.A.R.T reports: {{$labels.device}} ({{$labels.model}}) has errors."; + }; + + oom_kills = { + condition = "increase(kernel_vmstat_oom_kill[5m]) > 0"; + description = "{{$labels.instance}}: OOM kill detected"; + }; + + unusual_disk_read_latency = { + condition = "rate(diskio_read_time[1m]) / rate(diskio_reads[1m]) > 0.1 and rate(diskio_reads[1m]) > 0"; + description = "{{$labels.instance}}: Disk latency is growing (read operations > 100ms)\n"; + }; + + unusual_disk_write_latency = { + condition = "rate(diskio_write_time[1m]) / rate(diskio_write[1m]) > 0.1 and rate(diskio_write[1m]) > 0"; + description = "{{$labels.instance}}: Disk latency is growing (write operations > 100ms)\n"; + }; + + host_memory_under_memory_pressure = { + condition = "rate(node_vmstat_pgmajfault[1m]) > 1000"; + description = "{{$labels.instance}}: The node is under heavy memory pressure. High rate of major page faults: {{$value}}"; + }; + + ext4_errors = { + condition = "ext4_errors_value > 0"; + description = "{{$labels.instance}}: ext4 has reported {{$value}} I/O errors: check /sys/fs/ext4/*/errors_count"; + }; + + alerts_silences_changed = { + condition = ''abs(delta(alertmanager_silences{state="active"}[1h])) >= 1''; + description = "alertmanager: number of active silences has changed: {{$value}}"; + }; + }) diff --git a/lass/2configs/monitoring/prometheus.nix b/lass/2configs/monitoring/prometheus.nix new file mode 100644 index 00000000..ba32c62a --- /dev/null +++ b/lass/2configs/monitoring/prometheus.nix @@ -0,0 +1,110 @@ +{ config, lib, pkgs, ... }: +{ + #prometheus + krebs.iptables = { + enable = true; + tables.filter.INPUT.rules = [ + { predicate = "-i retiolum -p tcp --dport 80"; target = "ACCEPT"; } # nginx + # { predicate = "-i retiolum -p tcp --dport 3012"; target = "ACCEPT"; } # grafana + # { predicate = "-i retiolum -p tcp --dport 9093"; target = "ACCEPT"; } # alertmanager + # { predicate = "-i retiolum -p tcp --dport 9223"; target = "ACCEPT"; } # alertmanager + ]; + }; + + services.nginx = { + enable = true; + virtualHosts = { + "prometheus.lass.r" = { + locations."/".proxyPass = "http://localhost:9090"; + }; + "alert.lass.r" = { + locations."/".proxyPass = "http://localhost:9093"; + }; + "grafana.lass.r" = { + locations."/".proxyPass = "http://localhost:3012"; + }; + }; + }; + + services.grafana = { + enable = true; + addr = "0.0.0.0"; + port = 3012; + auth.anonymous = { + enable = true; + org_role = "Admin"; + }; + }; + services.prometheus = { + enable = true; + ruleFiles = [ + (pkgs.writeText "prometheus-rules.yml" (builtins.toJSON { + groups = [{ + name = "alerting-rules"; + rules = import ./alert-rules.nix { inherit lib; }; + }]; + })) + ]; + scrapeConfigs = [ + { + job_name = "telegraf"; + scrape_interval = "60s"; + metrics_path = "/metrics"; + static_configs = [ + { + targets = [ + "prism.r:9273" + "dishfire.r:9273" + "yellow.r:9273" + ]; + } + ]; + } + ]; + alertmanagers = [ + { scheme = "http"; + path_prefix = "/"; + static_configs = [ { targets = [ "localhost:9093" ]; } ]; + } + ]; + alertmanager = { + enable = true; + webExternalUrl = "https://alert.lass.r"; + listenAddress = "[::1]"; + configuration = { + global = { + # The smarthost and SMTP sender used for mail notifications. + smtp_smarthost = "localhost:587"; + smtp_from = "alertmanager@alert.lass.r"; + # smtp_auth_username = "alertmanager@thalheim.io"; + # smtp_auth_password = "$SMTP_PASSWORD"; + }; + route = { + receiver = "default"; + routes = [ + { + group_by = [ "host" ]; + group_wait = "30s"; + group_interval = "2m"; + repeat_interval = "2h"; + receiver = "all"; + } + ]; + }; + receivers = [ + { + name = "all"; + webhook_configs = [{ + url = "http://127.0.0.1:9223/"; + max_alerts = 5; + }]; + } + { + name = "default"; + } + ]; + }; + }; + }; + +} diff --git a/lass/2configs/monitoring/telegraf.nix b/lass/2configs/monitoring/telegraf.nix new file mode 100644 index 00000000..5258b87e --- /dev/null +++ b/lass/2configs/monitoring/telegraf.nix @@ -0,0 +1,72 @@ +{ config, lib, pkgs, ... }: +let + isVM = lib.any (mod: mod == "xen-blkfront" || mod == "virtio_console") config.boot.initrd.kernelModules; +in { + + krebs.iptables.tables.filter.INPUT.rules = [ + { predicate = "-i retiolum -p tcp --dport 9273"; target = "ACCEPT"; } + ]; + + systemd.services.telegraf.path = [ pkgs.nvme-cli ]; + + services.telegraf = { + enable = true; + extraConfig = { + agent.interval = "60s"; + inputs = { + http_response = [ + { urls = [ + "http://localhost:8080/about/health/" + ]; } + ]; + prometheus.metric_version = 2; + kernel_vmstat = { }; + # smart = lib.mkIf (!isVM) { + # path = pkgs.writeShellScript "smartctl" '' + # exec /run/wrappers/bin/sudo ${pkgs.smartmontools}/bin/smartctl "$@" + # ''; + # }; + system = { }; + mem = { }; + file = [{ + data_format = "influx"; + file_tag = "name"; + files = [ "/var/log/telegraf/*" ]; + }] ++ lib.optional (lib.any (fs: fs == "ext4") config.boot.supportedFilesystems) { + name_override = "ext4_errors"; + files = [ "/sys/fs/ext4/*/errors_count" ]; + data_format = "value"; + }; + exec = lib.optionalAttrs (lib.any (fs: fs == "zfs") config.boot.supportedFilesystems) { + ## Commands array + commands = [ + (pkgs.writeScript "zpool-health" '' + #!${pkgs.gawk}/bin/awk -f + BEGIN { + while ("${pkgs.zfs}/bin/zpool status" | getline) { + if ($1 ~ /pool:/) { printf "zpool_status,name=%s ", $2 } + if ($1 ~ /state:/) { printf " state=\"%s\",", $2 } + if ($1 ~ /errors:/) { + if (index($2, "No")) printf "errors=0i\n"; else printf "errors=%di\n", $2 + } + } + } + '') + ]; + data_format = "influx"; + }; + systemd_units = { }; + swap = { }; + disk.tagdrop = { + fstype = [ "tmpfs" "ramfs" "devtmpfs" "devfs" "iso9660" "overlay" "aufs" "squashfs" ]; + device = [ "rpc_pipefs" "lxcfs" "nsfs" "borgfs" ]; + }; + diskio = { }; + }; + outputs.prometheus_client = { + listen = ":9273"; + metric_version = 2; + }; + }; + }; +} -- cgit v1.2.3