generated from onedr0p/cluster-template
-
Notifications
You must be signed in to change notification settings - Fork 0
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
add democratic-csi local-path and loki
- Loading branch information
Showing
14 changed files
with
672 additions
and
2 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
|
|
@@ -6,3 +6,5 @@ resources: | |
| - speedtest-exporter | ||
| - thanos | ||
| - network-ups-tools/ks.yaml | ||
| - loki/secret.sops.yaml | ||
| - loki/ks.yaml | ||
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,206 @@ | ||
| --- | ||
| # yaml-language-server: $schema=https://raw.githubusercontent.com/fluxcd-community/flux2-schemas/main/helmrelease-helm-v2beta2.json | ||
| apiVersion: helm.toolkit.fluxcd.io/v2beta2 | ||
| kind: HelmRelease | ||
| metadata: | ||
| name: loki | ||
| spec: | ||
| interval: 30m | ||
| timeout: 15m | ||
| chart: | ||
| spec: | ||
| chart: loki | ||
| version: 5.41.4 | ||
| sourceRef: | ||
| kind: HelmRepository | ||
| name: grafana-charts | ||
| namespace: flux-system | ||
| interval: 30m | ||
| values: | ||
| loki: | ||
| structuredConfig: | ||
| auth_enabled: false | ||
|
|
||
| server: | ||
| log_level: info | ||
| http_listen_port: 3100 | ||
| grpc_listen_port: 9095 | ||
|
|
||
| grpc_server_max_recv_msg_size: 8388608 | ||
| grpc_server_max_send_msg_size: 8388608 | ||
|
|
||
| memberlist: | ||
| join_members: ["loki-memberlist"] | ||
|
|
||
| limits_config: | ||
| retention_period: 14d | ||
| enforce_metric_name: false | ||
| reject_old_samples: true | ||
| reject_old_samples_max_age: 168h | ||
| max_cache_freshness_per_query: 10m | ||
| split_queries_by_interval: 15m | ||
| ingestion_rate_mb: 50 | ||
| ingestion_burst_size_mb: 1000 | ||
| per_stream_rate_limit: 5M | ||
| per_stream_rate_limit_burst: 20M | ||
| shard_streams: | ||
| enabled: true | ||
|
|
||
| schema_config: | ||
| configs: | ||
| - from: "2021-08-01" | ||
| store: boltdb-shipper | ||
| object_store: s3 | ||
| schema: v12 | ||
| index: | ||
| prefix: loki_index_ | ||
| period: 24h | ||
|
|
||
| common: | ||
| path_prefix: /var/loki | ||
| replication_factor: 2 | ||
| storage: | ||
| s3: | ||
| s3: null | ||
| insecure: true | ||
| s3forcepathstyle: true | ||
| bucketnames: loki | ||
| endpoint: ${S3_BUCKET_HOST} | ||
| region: ${S3_BUCKET_REGION} | ||
| access_key_id: ${S3_ACCESS_KEY} | ||
| secret_access_key: ${S3_SECRET_KEY} | ||
| ring: | ||
| kvstore: | ||
| store: memberlist | ||
|
|
||
| ruler: | ||
| enable_api: true | ||
| enable_alertmanager_v2: true | ||
| alertmanager_url: http://prometheus-alertmanager.monitoring.svc.cluster.local:9093 | ||
| storage: | ||
| type: local | ||
| local: | ||
| directory: /rules | ||
| # rule_path: /rules | ||
| rule_path: /tmp/scratch | ||
| ring: | ||
| kvstore: | ||
| store: memberlist | ||
|
|
||
| distributor: | ||
| ring: | ||
| kvstore: | ||
| store: memberlist | ||
|
|
||
| compactor: | ||
| working_directory: /var/loki/boltdb-shipper-compactor | ||
| shared_store: s3 | ||
| compaction_interval: 10m | ||
| retention_enabled: true | ||
| retention_delete_delay: 2h | ||
| retention_delete_worker_count: 150 | ||
|
|
||
| ingester: | ||
| max_chunk_age: 1h | ||
| lifecycler: | ||
| ring: | ||
| kvstore: | ||
| store: memberlist | ||
|
|
||
| analytics: | ||
| reporting_enabled: false | ||
|
|
||
| podAnnotations: | ||
| secret.reloader.stakater.com/reload: loki-secret | ||
|
|
||
| gateway: | ||
| replicas: 2 | ||
| affinity: | | ||
| podAntiAffinity: | ||
| preferredDuringSchedulingIgnoredDuringExecution: | ||
| - weight: 1 | ||
| podAffinityTerm: | ||
| labelSelector: | ||
| matchLabels: | ||
| {{- include "loki.gatewaySelectorLabels" . | nindent 12 }} | ||
| topologyKey: kubernetes.io/hostname | ||
| enabled: true | ||
| image: | ||
| registry: ghcr.io | ||
| repository: nginxinc/nginx-unprivileged | ||
| tag: 1.25-alpine | ||
| ingress: | ||
| enabled: false | ||
|
|
||
| read: | ||
| replicas: 2 | ||
| affinity: | | ||
| podAntiAffinity: | ||
| preferredDuringSchedulingIgnoredDuringExecution: | ||
| - weight: 1 | ||
| podAffinityTerm: | ||
| labelSelector: | ||
| matchLabels: | ||
| {{- include "loki.readSelectorLabels" . | nindent 12 }} | ||
| topologyKey: kubernetes.io/hostname | ||
| write: | ||
| replicas: 2 | ||
| affinity: | | ||
| podAntiAffinity: | ||
| preferredDuringSchedulingIgnoredDuringExecution: | ||
| - weight: 1 | ||
| podAffinityTerm: | ||
| labelSelector: | ||
| matchLabels: | ||
| {{- include "loki.writeSelectorLabels" . | nindent 12 }} | ||
| topologyKey: kubernetes.io/hostname | ||
| persistence: | ||
| size: 20Gi | ||
| storageClass: local-hostpath | ||
|
|
||
| backend: | ||
| replicas: 2 | ||
| affinity: | | ||
| podAntiAffinity: | ||
| preferredDuringSchedulingIgnoredDuringExecution: | ||
| - weight: 1 | ||
| podAffinityTerm: | ||
| labelSelector: | ||
| matchLabels: | ||
| {{- include "loki.backendSelectorLabels" . | nindent 12 }} | ||
| topologyKey: kubernetes.io/hostname | ||
| persistence: | ||
| size: 20Gi | ||
| storageClass: local-hostpath | ||
| # extraVolumeMounts: | ||
| # - name: rules | ||
| # mountPath: /rules/fake | ||
| # - name: scratch | ||
| # mountPath: /tmp/scratch | ||
| # extraVolumes: | ||
| # - name: rules | ||
| # configMap: | ||
| # name: loki-alerting-rules | ||
| # - name: scratch | ||
| # emptyDir: {} | ||
|
|
||
| monitoring: | ||
| dashboards: | ||
| annotations: | ||
| grafana_folder: Loki | ||
| serviceMonitor: | ||
| enabled: false | ||
| metricsInstance: | ||
| enabled: false | ||
| selfMonitoring: | ||
| enabled: false | ||
| grafanaAgent: | ||
| installOperator: false | ||
| lokiCanary: | ||
| enabled: false | ||
| sidecar: | ||
| image: | ||
| repository: ghcr.io/kiwigrid/k8s-sidecar | ||
| test: | ||
| enabled: false |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,7 @@ | ||
| --- | ||
| # yaml-language-server: $schema=https://json.schemastore.org/kustomization | ||
| apiVersion: kustomize.config.k8s.io/v1beta1 | ||
| kind: Kustomization | ||
| resources: | ||
| - ./helm-release.yaml | ||
| - ./prometheus-rule.yaml |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,174 @@ | ||
| --- | ||
| # yaml-language-server: $schema=https://raw.githubusercontent.com/datreeio/CRDs-catalog/main/monitoring.coreos.com/prometheusrule_v1.json | ||
| apiVersion: monitoring.coreos.com/v1 | ||
| kind: PrometheusRule | ||
| metadata: | ||
| name: loki.rules | ||
| spec: | ||
| groups: | ||
| - name: loki.rules | ||
| rules: | ||
| - alert: LokiRequestErrors | ||
| annotations: | ||
| message: "{{ $labels.job }} {{ $labels.route }} is experiencing {{ $value | humanizePercentage }} errors." | ||
| expr: | | ||
| 100 * sum(rate(loki_request_duration_seconds_count{status_code=~"5.."}[1m])) by (namespace, job, route) | ||
| / | ||
| sum(rate(loki_request_duration_seconds_count[1m])) by (namespace, job, route) | ||
| > 10 | ||
| for: 15m | ||
| labels: | ||
| severity: critical | ||
| - alert: LokiRequestPanics | ||
| annotations: | ||
| message: "{{ $labels.job }} is experiencing {{ $value | humanizePercentage }} increase of panics." | ||
| expr: | | ||
| sum(increase(loki_panic_total[10m])) by (namespace, job) | ||
| > 0 | ||
| labels: | ||
| severity: critical | ||
| - alert: LokiRequestLatency | ||
| annotations: | ||
| message: "{{ $labels.job }} {{ $labels.route }} is experiencing {{ $value }}s 99th percentile latency." | ||
| expr: | | ||
| namespace_job_route:loki_request_duration_seconds:99quantile{route!~"(?i).*tail.*"} | ||
| > 1 | ||
| for: 15m | ||
| labels: | ||
| severity: critical | ||
| - expr: | | ||
| histogram_quantile(0.99, sum(rate(loki_request_duration_seconds_bucket[1m])) | ||
| by (le, job)) | ||
| record: job:loki_request_duration_seconds:99quantile | ||
| - expr: | | ||
| histogram_quantile(0.50, sum(rate(loki_request_duration_seconds_bucket[1m])) | ||
| by (le, job)) | ||
| record: job:loki_request_duration_seconds:50quantile | ||
| - expr: | | ||
| sum(rate(loki_request_duration_seconds_sum[1m])) by (job) | ||
| / | ||
| sum(rate(loki_request_duration_seconds_count[1m])) by (job) | ||
| record: job:loki_request_duration_seconds:avg | ||
| - expr: | | ||
| sum(rate(loki_request_duration_seconds_bucket[1m])) | ||
| by (le, job) | ||
| record: job:loki_request_duration_seconds_bucket:sum_rate | ||
| - expr: | | ||
| sum(rate(loki_request_duration_seconds_sum[1m])) by (job) | ||
| record: job:loki_request_duration_seconds_sum:sum_rate | ||
| - expr: | | ||
| sum(rate(loki_request_duration_seconds_count[1m])) by (job) | ||
| record: job:loki_request_duration_seconds_count:sum_rate | ||
| - expr: | | ||
| histogram_quantile(0.99, sum(rate(loki_request_duration_seconds_bucket[1m])) | ||
| by (le, job, route)) | ||
| record: job_route:loki_request_duration_seconds:99quantile | ||
| - expr: | | ||
| histogram_quantile(0.50, sum(rate(loki_request_duration_seconds_bucket[1m])) | ||
| by (le, job, route)) | ||
| record: job_route:loki_request_duration_seconds:50quantile | ||
| - expr: | | ||
| sum(rate(loki_request_duration_seconds_sum[1m])) by (job, route) | ||
| / | ||
| sum(rate(loki_request_duration_seconds_count[1m])) by (job, route) | ||
| record: job_route:loki_request_duration_seconds:avg | ||
| - expr: | | ||
| sum(rate(loki_request_duration_seconds_bucket[1m])) | ||
| by (le, job, route) | ||
| record: job_route:loki_request_duration_seconds_bucket:sum_rate | ||
| - expr: | | ||
| sum(rate(loki_request_duration_seconds_sum[1m])) by (job, route) | ||
| record: job_route:loki_request_duration_seconds_sum:sum_rate | ||
| - expr: | | ||
| sum(rate(loki_request_duration_seconds_count[1m])) by (job, route) | ||
| record: job_route:loki_request_duration_seconds_count:sum_rate | ||
| - expr: | | ||
| histogram_quantile(0.99, sum(rate(loki_request_duration_seconds_bucket[1m])) | ||
| by (le, namespace, job, route)) | ||
| record: namespace_job_route:loki_request_duration_seconds:99quantile | ||
| - expr: | | ||
| histogram_quantile(0.50, sum(rate(loki_request_duration_seconds_bucket[1m])) | ||
| by (le, namespace, job, route)) | ||
| record: namespace_job_route:loki_request_duration_seconds:50quantile | ||
| - expr: | | ||
| sum(rate(loki_request_duration_seconds_sum[1m])) by (namespace, job, route) | ||
| / | ||
| sum(rate(loki_request_duration_seconds_count[1m])) by (namespace, job, route) | ||
| record: namespace_job_route:loki_request_duration_seconds:avg | ||
| - expr: | | ||
| sum(rate(loki_request_duration_seconds_bucket[1m])) | ||
| by (le, namespace, job, route) | ||
| record: namespace_job_route:loki_request_duration_seconds_bucket:sum_rate | ||
| - expr: | | ||
| sum(rate(loki_request_duration_seconds_sum[1m])) | ||
| by (namespace, job, route) | ||
| record: namespace_job_route:loki_request_duration_seconds_sum:sum_rate | ||
| - expr: | | ||
| sum(rate(loki_request_duration_seconds_count[1m])) | ||
| by (namespace, job, route) | ||
| record: namespace_job_route:loki_request_duration_seconds_count:sum_rate | ||
| - name: smart | ||
| rules: | ||
| - alert: SMARTFailure | ||
| expr: | | ||
| sum by (hostname) (count_over_time({hostname=~".+"} | json | _SYSTEMD_UNIT = "smartmontools.service" !~ "(?i)previous self-test completed without error" !~ "(?i)Prefailure" |~ "(?i)(error|fail)"[2m])) > 0 | ||
| for: 2m | ||
| labels: | ||
| severity: critical | ||
| category: logs | ||
| annotations: | ||
| hostname: "{{ $labels.hostname }}" | ||
| summary: "{{ $labels.hostname }} has reported SMART failures" | ||
|
|
||
| - name: frigate | ||
| rules: | ||
| - alert: FrigateMQTTUnreachable | ||
| expr: | | ||
| sum(count_over_time({app="frigate"} |~ "(?i)unable to connect to mqtt server"[2m])) > 0 | ||
| for: 2m | ||
| labels: | ||
| severity: critical | ||
| category: logs | ||
| annotations: | ||
| app: "{{ $labels.app }}" | ||
| summary: "{{ $labels.app }} is unable to reach MQTT" | ||
|
|
||
| - name: home-assistant | ||
| rules: | ||
| - alert: HomeAssistantPostgresUnreachable | ||
| expr: | | ||
| sum by (app) (count_over_time({app="home-assistant"} |~ "(?i)error in database connectivity"[2m])) > 0 | ||
| for: 2m | ||
| labels: | ||
| severity: critical | ||
| category: logs | ||
| annotations: | ||
| app: "{{ $labels.app }}" | ||
| summary: "{{ $labels.app }} is unable to connect to postgres" | ||
|
|
||
| # - name: zwave-js-ui | ||
| # rules: | ||
| # - alert: ZwaveMQTTUnreachable | ||
| # expr: | | ||
| # sum(count_over_time({app="zwave-js-ui"} |~ "(?i)error while connecting mqtt"[2m])) > 0 | ||
| # for: 2m | ||
| # labels: | ||
| # severity: critical | ||
| # category: logs | ||
| # annotations: | ||
| # app: "{{ $labels.app }}" | ||
| # summary: "{{ $labels.app }} is unable to reach MQTT" | ||
|
|
||
| # - name: zigbee2mqtt | ||
| # rules: | ||
| # - alert: ZigbeeMQTTUnreachable | ||
| # expr: | | ||
| # sum(count_over_time({app="zigbee2mqtt"} |~ "(?i)not connected to mqtt server"[2m])) > 0 | ||
| # for: 2m | ||
| # labels: | ||
| # severity: critical | ||
| # category: logs | ||
| # annotations: | ||
| # app: "{{ $labels.app }}" | ||
| # summary: "{{ $labels.app }} is unable to reach MQTT" |
Oops, something went wrong.