change folder to monitoring

Reviewed-on: #35
Reviewed-by: Danny Groenewegen <247+dannygroenewegen@noreply.git.coopcloud.tech>
Reviewed-by: moritz <98+moritz@noreply.git.coopcloud.tech>
Co-authored-by: f <f@sutty.nl>

.env.sample

@@ -20,6 +20,9 @@ SECRET_BASIC_AUTH_VERSION=v1
 # server is remote
 # PROMETHEUS_REMOTE_WRITE_URL=https://prometheus.$DOMAIN/api/v1/write
 
+# Enable container filesystem usage metrics (expensive du-scan, ~30% CPU on large hosts)
+# CADVISOR_DISK_USAGE=1
+
 # Enable authenticated scraping of containers that opt in via
 # prometheus.io/auth=basic or prometheus.io/auth=bearer labels (used as
 # password/bearer token respectively). Insert it with:
@@ -50,6 +53,10 @@ SECRET_BASIC_AUTH_VERSION=v1
 # Not needed if you just want to read local log files — use SYSLOG_FILES instead.
 # SYSLOG=1
 # COMPOSE_FILE="$COMPOSE_FILE:compose.syslog.yml"
+#
+# Scrape Docker-Events via official docker-cli image using docker socket
+# COMPOSE_FILE="$COMPOSE_FILE:compose.docker-events.yml"
+
 
 # Monitoring Server
 #
@@ -65,6 +72,8 @@ SECRET_BASIC_AUTH_VERSION=v1
 #
 # COMPOSE_FILE="$COMPOSE_FILE:compose.loki.yml"
 #
+# LOKI_LOG_LEVEL=warn
+#
 # Set to 0 to disable retention
 # LOKI_RETENTION_PERIOD=744h
 # LOKI_STORAGE_FILESYSTEM=1
@@ -84,7 +93,7 @@ SECRET_BASIC_AUTH_VERSION=v1
 # SECRET_GF_ADMINPASSWD_VERSION=v1
 ## Grafana's own domain. Defaults to $DOMAIN
 ## Change the value if you want Grafana on another domain.
-GRAFANA_DOMAIN=$DOMAIN
+# GRAFANA_DOMAIN=$DOMAIN
 #
 ## Single-Sign-On with OIDC
 # COMPOSE_FILE="$COMPOSE_FILE:compose.grafana-oidc.yml"
@@ -123,3 +132,12 @@ GRAFANA_DOMAIN=$DOMAIN
 
 # Node memory usage alert will trigger when memory usage is above the given number in percent
 #ALERT_NODE_MEMORY_USAGE=85
+
+# Loki ingestion alert will trigger when a host ships fewer than the given number of log entries per second
+#ALERT_LOKI_INGESTION_STALLED=0.01
+
+# Loki write error alert will trigger when log pushes fail at more than the given rate per second
+#ALERT_LOKI_WRITE_ERRORS=0
+
+# Loki WAL disk alert will trigger when the filesystem holding the WAL is above the given number in percent. Loki throttles all writes at 90
+#ALERT_LOKI_WAL_DISK_USAGE=85

.gitea/CODEOWNERS

@@ -0,0 +1 @@
+* @coop-cloud/monitoring-ng-maintainers

MAINTENANCE.md

@@ -0,0 +1,35 @@
+# Monitoring NG Recipe Maintenance
+
+All contributions should be made via a pull request. This is to ensure a
+certain quality and consistency, that others can rely on.
+
+## Maintainer Responsibilities
+
+A recipe maintainer has the following responsibilities:
+
+- Respond to pull requests / issues within two weeks
+- Make image security updates within a week
+- Make image major updates every three months
+
+In order to fullfill these responsibilities a recipe maintainer:
+
+- Has to watch the repository (to get notifications)
+- Needs to make sure renovate is configured properly
+
+## Pull Requests
+
+A pull request can be merged if it is approved by at least one maintainer. For
+pull requests opened by a maintainer they need to be approved by another
+maintainer. Even though it is okay to merge a pull request with one approval, it
+is always better if all maintainers looked at the pull request and approved it.
+
+## Become a maintainer
+
+Everyone can apply to be a recipe maintainer:
+1. Watch the repository to always get updates
+2. Simply add your self to the list in the [README.md](./README.md) and open a new pull request with the change.
+3. Once the pull request gets merged you will be added to the
+   [monitoring-ng maintainers
+   team](https://git.coopcloud.tech/org/coop-cloud/teams/monitoring-ng-maintainers).
+4. Join the room
+   [#cc-|-monitoring-ng-maintenance:matrix.org](#cc-|-monitoring-ng-maintenance:matrix.org) and chat to other maintainers.

README.md

@@ -7,7 +7,7 @@ It's based heavily on the [monitoring-lite](https://git.coopcloud.tech/coop-clou
 Metrics and logs are collected by [Grafana Alloy](https://grafana.com/docs/alloy/latest/) and pushed to a central Prometheus/Loki (via `remote_write`/`loki push`). Every `monitoring-ng` instance runs its own Alloy, whether or not it also runs the central stack.
 
 <!-- metadata -->
-
+- **Maintainer**: [@coop-cloud/monitoring-ng-maintainers](https://git.coopcloud.tech/org/coop-cloud/teams/monitoring-ng-maintainers)
 - **Category**: Apps
 - **Status**: 2, beta
 - **Image**: [`grafana/grafana`](https://hub.docker.com/r/grafana/grafana), 4, upstream
@@ -54,6 +54,14 @@ This is what a gathering host pushes into. It also runs its own Alloy, so it mon
 
 ## Additional features
 
+### Scrape docker events
+
+Just uncomment the following line in your .env to collect docker events using
+the official docker-cli image and the docker socket:
+```
+COMPOSE_FILE="$COMPOSE_FILE:compose.docker-events.yml"
+```
+
 ### Discovering metrics from other apps
 
 Alloy auto-discovers and scrapes other Docker Swarm services running on the same host, on the `proxy` network, that opt in via labels. No manual scrape config needed. On the app's `compose.yml`:

abra.sh

@@ -5,12 +5,13 @@ export GF_STACKS_DASH_VERSION=v3
 export GF_TRAEFIK_DASH_VERSION=v3
 export GF_BACKUP_DASH_VERSION=v1
 export GF_CUSTOM_INI_VERSION=v4
-export LOKI_YML_VERSION=v3
+export LOKI_YML_VERSION=v4
 export PROMETHEUS_YML_VERSION=v2
 export MATRIX_ALERTMANAGER_CONFIG_VERSION=v1
 export MATRIX_ALERTMANAGER_ENTRYPOINT_VERSION=v1
 export GF_ALERTS_NODE_VERSION=v3
-export CONFIG_ALLOY_VERSION=v1
+export GF_ALERTS_LOKI_VERSION=v5
+export CONFIG_ALLOY_VERSION=v3
 
 # migrates secrets from old names to new names by reading values from the
 # running containers on the server and re-inserting them under the new names.

alerts/loki.yml.tmpl

@@ -0,0 +1,191 @@
+apiVersion: 1
+
+# List of alert rule UIDs that should be deleted
+deleteRules:
+  {{ if not (env "ALERT_LOKI_INGESTION_STALLED") }}
+  - orgId: 1
+    uid: coopcloud_loki_ingestion_stalled
+  {{ end }}
+  {{ if not (env "ALERT_LOKI_WRITE_ERRORS") }}
+  - orgId: 1
+    uid: coopcloud_loki_write_errors
+  {{ end }}
+  {{ if not (env "ALERT_LOKI_WAL_DISK_USAGE") }}
+  - orgId: 1
+    uid: coopcloud_loki_wal_disk_usage
+  {{ end }}
+
+groups:
+- orgId: 1
+  name: loki
+  folder: monitoring
+  interval: 5m
+  rules:
+    {{ if (env "ALERT_LOKI_INGESTION_STALLED") }}
+    - uid: coopcloud_loki_ingestion_stalled
+      title: Loki Ingestion Stalled
+      condition: C
+      data:
+        - refId: A
+          relativeTimeRange:
+            from: 1800
+            to: 0
+          datasourceUid: PBFA97CFB590B2093
+          model:
+            editorMode: code
+            expr: sum by (instance) (rate(loki_write_sent_entries_total[15m]))
+            instant: true
+            intervalMs: 1000
+            legendFormat: __auto
+            maxDataPoints: 43200
+            range: false
+            refId: A
+        - refId: C
+          relativeTimeRange:
+            from: 1800
+            to: 0
+          datasourceUid: __expr__
+          model:
+            conditions:
+                - evaluator:
+                    params:
+                        - {{ env "ALERT_LOKI_INGESTION_STALLED" }}
+                    type: lt
+                  operator:
+                    type: and
+                  query:
+                    params:
+                        - C
+                  reducer:
+                    params: []
+                    type: last
+                  type: query
+            datasource:
+                type: __expr__
+                uid: __expr__
+            expression: A
+            intervalMs: 1000
+            maxDataPoints: 43200
+            refId: C
+            type: threshold
+      noDataState: Alerting
+      execErrState: Error
+      for: 15m
+      annotations:
+        description: ""
+        runbook_url: ""
+        summary: Loki received no logs from {{`{{ $labels.instance }}`}} for 15m ({{`{{ printf "%.4f" (index $values "A").Value }}`}} entries/sec, threshold {{ env "ALERT_LOKI_INGESTION_STALLED" }})
+      isPaused: false
+    {{ end }}
+    {{ if (env "ALERT_LOKI_WRITE_ERRORS") }}
+    - uid: coopcloud_loki_write_errors
+      title: Loki Write Errors
+      condition: C
+      data:
+        - refId: A
+          relativeTimeRange:
+            from: 1200
+            to: 0
+          datasourceUid: PBFA97CFB590B2093
+          model:
+            editorMode: code
+            expr: sum by (instance, status_code) (rate(loki_write_request_duration_seconds_count{status_code!~"2.."}[10m]))
+            instant: true
+            intervalMs: 1000
+            legendFormat: __auto
+            maxDataPoints: 43200
+            range: false
+            refId: A
+        - refId: C
+          relativeTimeRange:
+            from: 1200
+            to: 0
+          datasourceUid: __expr__
+          model:
+            conditions:
+                - evaluator:
+                    params:
+                        - {{ env "ALERT_LOKI_WRITE_ERRORS" }}
+                    type: gt
+                  operator:
+                    type: and
+                  query:
+                    params:
+                        - C
+                  reducer:
+                    params: []
+                    type: last
+                  type: query
+            datasource:
+                type: __expr__
+                uid: __expr__
+            expression: A
+            intervalMs: 1000
+            maxDataPoints: 43200
+            refId: C
+            type: threshold
+      noDataState: NoData
+      execErrState: Error
+      for: 10m
+      annotations:
+        description: ""
+        runbook_url: ""
+        summary: Loki is rejecting log pushes from {{`{{ $labels.instance }}`}} with HTTP {{`{{ $labels.status_code }}`}} ({{`{{ printf "%.2f" (index $values "A").Value }}`}} failed pushes/sec)
+      isPaused: false
+    {{ end }}
+    {{ if (env "ALERT_LOKI_WAL_DISK_USAGE") }}
+    - uid: coopcloud_loki_wal_disk_usage
+      title: Loki WAL Disk Usage
+      condition: C
+      data:
+        - refId: A
+          relativeTimeRange:
+            from: 600
+            to: 0
+          datasourceUid: PBFA97CFB590B2093
+          model:
+            editorMode: code
+            expr: max by (instance) (loki_ingester_wal_disk_usage_percent * 100)
+            instant: true
+            intervalMs: 1000
+            legendFormat: __auto
+            maxDataPoints: 43200
+            range: false
+            refId: A
+        - refId: C
+          relativeTimeRange:
+            from: 600
+            to: 0
+          datasourceUid: __expr__
+          model:
+            conditions:
+                - evaluator:
+                    params:
+                        - {{ env "ALERT_LOKI_WAL_DISK_USAGE" }}
+                    type: gt
+                  operator:
+                    type: and
+                  query:
+                    params:
+                        - C
+                  reducer:
+                    params: []
+                    type: last
+                  type: query
+            datasource:
+                type: __expr__
+                uid: __expr__
+            expression: A
+            intervalMs: 1000
+            maxDataPoints: 43200
+            refId: C
+            type: threshold
+      noDataState: NoData
+      execErrState: Error
+      for: 10m
+      annotations:
+        description: ""
+        runbook_url: ""
+        summary: Loki WAL disk {{`{{ printf "%.1f" (index $values "A").Value }}`}}% full on {{`{{ $labels.instance }}`}} (threshold {{ env "ALERT_LOKI_WAL_DISK_USAGE" }}%, writes stop at 90%)
+      isPaused: false
+    {{ end }}

compose.docker-events.yml

@@ -0,0 +1,20 @@
+version: "3.8"
+services:
+  docker-events:
+    image: docker:29.8.1-cli
+    command: >
+      docker events
+      --format '{{json .}}'
+      --filter type=container
+      --filter event=start
+      --filter event=die
+      --filter event=kill
+      --filter event=oom
+      --filter event=restart
+      --filter event=stop
+      --filter event=health_status
+    volumes:
+      - /var/run/docker.sock:/var/run/docker.sock:ro
+    deploy:
+      restart_policy:
+        condition: on-failure

compose.grafana.yml

@@ -24,6 +24,8 @@ services:
         target: /var/lib/grafana/dashboards/backup.json
       - source: gf_alerts_node
         target: /etc/grafana/provisioning/alerting/node.yml
+      - source: gf_alerts_loki
+        target: /etc/grafana/provisioning/alerting/loki.yml
     networks:
       - proxy
       - internal
@@ -34,6 +36,9 @@ services:
       - GF_INSTALL_PLUGINS
       - ALERT_NODE_DISK_SPACE_LEFT
       - ALERT_NODE_MEMORY_USAGE
+      - ALERT_LOKI_INGESTION_STALLED
+      - ALERT_LOKI_WRITE_ERRORS
+      - ALERT_LOKI_WAL_DISK_USAGE
     deploy:
       labels:
         - "traefik.enable=true"
@@ -77,6 +82,10 @@ configs:
     template_driver: golang
     name: ${STACK_NAME}_gf_alerts_node_${GF_ALERTS_NODE_VERSION}
     file: alerts/node.yml.tmpl
+  gf_alerts_loki:
+    template_driver: golang
+    name: ${STACK_NAME}_gf_alerts_loki_${GF_ALERTS_LOKI_VERSION}
+    file: alerts/loki.yml.tmpl
 
 volumes:
   grafana-data:

compose.loki.yml

@@ -19,9 +19,16 @@ services:
       - LOKI_AWS_REGION
       - LOKI_BUCKET_NAMES
       - LOKI_RETENTION_PERIOD
+      - LOKI_LOG_LEVEL
       - STACK_NAME
       - LOKI_STORAGE_FILESYSTEM
       - LOKI_STORAGE_S3
+    healthcheck:
+      test: ["CMD", "/usr/bin/loki", "-health"]
+      interval: 10s
+      timeout: 10s
+      retries: 10
+      start_period: 30s
     deploy:
       restart_policy:
         condition: on-failure
@@ -34,6 +41,8 @@ services:
         - "traefik.http.routers.${STACK_NAME}-loki.tls=true"
         - "traefik.http.routers.${STACK_NAME}-loki.tls.certresolver=${LETS_ENCRYPT_ENV}"
         - "traefik.http.routers.${STACK_NAME}-loki.middlewares=basicauth@file"
+        - "prometheus.io/scrape=true"
+        - "prometheus.io/port=3100"
 
 
 configs:

compose.yml

@@ -46,14 +46,14 @@ services:
       test: ["CMD", "bash", "-c", "exec 3<>/dev/tcp/localhost/12345 && printf 'GET /-/ready HTTP/1.0\r\nHost: localhost\r\n\r\n' >&3 && head -1 <&3 | grep -q 200"]
       interval: 5s
       timeout: 10s
-      retries: 3
-      start_period: 10s
+      retries: 10
+      start_period: 30s
     deploy:
       restart_policy:
         condition: on-failure
       labels:
         - "backupbot.backup=${ENABLE_BACKUPS:-true}"
-        - "coop-cloud.${STACK_NAME}.version=1.6.0+v1.8.1"
+        - "coop-cloud.${STACK_NAME}.version=2.2.0+v1.18.1"
 configs:
   config_alloy:
     template_driver: golang

config.alloy.tmpl

@@ -14,7 +14,11 @@ discovery.docker "linux" {
 {{ if ne (env "PROMETHEUS_REMOTE_WRITE_URL") "" }}
 prometheus.exporter.cadvisor "docker" {
   docker_only = true
+  {{ if eq (env "CADVISOR_DISK_USAGE") "1" }}
   enabled_metrics = ["cpu", "cpuLoad", "disk", "diskIO", "memory", "network"]
+  {{ else }}
+  enabled_metrics = ["cpu", "cpuLoad", "diskIO", "memory", "network"]
+  {{ end }}
   // host-wide totals already come from prometheus.exporter.unix
   disable_root_cgroup_stats = true
 }
@@ -48,7 +52,32 @@ prometheus.scrape "default" {
     prometheus.exporter.cadvisor.docker.targets,
   )
 
+  forward_to = [prometheus.relabel.container_meta.receiver]
+}
+
+prometheus.relabel "container_meta" {
   forward_to = [prometheus.remote_write.prometheus.receiver]
+
+  // remove sha tail: nginx:1.31.1@sha256:608a... -> nginx:1.31.1
+  rule {
+    source_labels = ["image"]
+    regex         = "([^@]+)@sha256:.*"
+    target_label  = "image"
+    replacement   = "$1"
+  }
+  // split image in name and tag
+  rule {
+    source_labels = ["image"]
+    regex         = "(.+):[^:/]+"
+    target_label  = "image_name"
+    replacement   = "$1"
+  }
+  rule {
+    source_labels = ["image"]
+    regex         = ".+:([^:/]+)"
+    target_label  = "image_tag"
+    replacement   = "$1"
+  }
 }
 
 prometheus.remote_write "prometheus" {
@@ -286,7 +315,35 @@ loki.source.docker "docker" {
 loki.source.journal "journal" {
   path = "/rootfs/var/log/journal"
   labels = { job = "{{ env "DOMAIN" }}" }
+  relabel_rules = loki.relabel.journal.rules
+  forward_to = [loki.process.journal.receiver]
+}
+
+loki.relabel "journal" {
+  forward_to = []
+
+  rule {
+    source_labels = ["__journal__systemd_unit"]
+    target_label  = "unit"
+  }
+  rule {
+    source_labels = ["__journal_syslog_identifier"]
+    target_label  = "ident"
+  }
+  rule {
+    source_labels = ["__journal_priority_keyword"]
+    target_label  = "priority"
+  }
+}
+
+// drop network db stats (15-20% of kernel logs)
+loki.process "journal" {
   forward_to = [loki.write.loki.receiver]
+
+  stage.drop {
+    expression          = ".*NetworkDB stats.*"
+    drop_counter_reason = "networkdb_stats"
+  }
 }
 {{ end }}
 

loki.yml.tmpl

@@ -15,7 +15,7 @@ ruler:
 
 server:
   http_listen_port: 3100
-  log_level: warn
+  log_level: {{ or (env "LOKI_LOG_LEVEL") "warn" }}
 
 distributor:
   ring:
@@ -127,4 +127,4 @@ frontend:
 
 table_manager:
   retention_deletes_enabled: false
-  retention_period: 0s
+  retention_period: 0s

release/2.0.0+v1.18.1

@@ -59,3 +59,6 @@ See the README's "Auto-discovering metrics from other apps" section.
 The Swarm, Stacks and Traefik dashboards were reworked to show old (pull-model)
 and new (Alloy push-model) data as one continuous line, so you don't lose history
 across the migration.
+
+The disk usage per container was disabled per default because of the CPU-expensive 
+filesystem scan (30% cpu increase). Enable by uncommenting env CADVISOR_DISK_USAGE=1

release/2.1.0+v1.18.1

@@ -0,0 +1 @@
+alloy: split image label in "image_name" and "image_tag", relabel journald logs and drop NetworkDB stats

release/2.2.0+v1.18.1

@@ -0,0 +1 @@
+allow scraping of docker events via new compose.docker-events.yml