.env.sample

@@ -72,6 +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
@@ -130,3 +132,12 @@ SECRET_BASIC_AUTH_VERSION=v1
 
 # 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

abra.sh

@@ -5,11 +5,12 @@ 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 GF_ALERTS_LOKI_VERSION=v5
 export CONFIG_ALLOY_VERSION=v3
 
 # migrates secrets from old names to new names by reading values from the

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.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:

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