Add Grafana Loki integration (#2156)

Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Co-authored-by: Vinicius D. Cerutti <[email protected]>

Author: 3 people

.cirun.yml

@@ -4,8 +4,8 @@ runners:
   - name: run-k8s-tests
     # Cloud Provider: AWS
     cloud: aws
-    # Instance Type has 4 vcpu, 16 GiB memory, Up to 5 Gbps Network Performance
-    instance_type: t3a.xlarge
+    # Instance Type has 8 vcpu, 32 GiB memory, Up to 5 Gbps Network Performance
+    instance_type: t3a.2xlarge
     # Custom AMI with docker/cypress/hub pre-installed
     machine_image: ami-0a388df278199ff52
     # Region: Oregon

.github/workflows/test_local_integration.yaml

@@ -96,7 +96,6 @@ jobs:
           sed -i -E 's/(cpu_guarantee):\s+[0-9\.]+/\1: 0.25/g' "nebari-config.yaml"
           sed -i -E 's/(mem_guarantee):\s+[A-Za-z0-9\.]+/\1: 0.25G/g' "nebari-config.yaml"
 
-
           # Change default JupyterLab theme
           cat >> nebari-config.yaml <<- EOM
           jupyterlab:
@@ -105,6 +104,16 @@ jobs:
                 theme: JupyterLab Dark
           EOM
 
+          # Change default value for minio persistence size
+          cat >> nebari-config.yaml <<- EOM
+          monitoring:
+            enabled: true
+            overrides:
+              minio:
+                persistence:
+                  size: 1Gi
+          EOM
+
           cat nebari-config.yaml
 
       - name: Deploy Nebari
@@ -115,7 +124,7 @@ jobs:
       - name: Basic kubectl checks after deployment
         if: always()
         run: |
-          kubectl get all,cm,secret,ing -A
+          kubectl get all,cm,secret,pv,pvc,ing -A
 
       - name: Check github-actions.nebari.dev resolves
         run: |

RELEASE.md

@@ -11,6 +11,8 @@ This file is copied to nebari-dev/nebari-docs using a GitHub Action. -->
 
 ## Upcoming Release
 
+* Added Grafana Loki to aggregate, index and search logs
+
 ## Release 2024.1.1 - January 17, 2024
 
 ### Feature changes and enhancements

src/_nebari/stages/kubernetes_services/__init__.py

@@ -199,8 +199,16 @@ class JHubApps(schema.Base):
     enabled: bool = False
 
 
+class MonitoringOverrides(schema.Base):
+    loki: typing.Dict = {}
+    promtail: typing.Dict = {}
+    minio: typing.Dict = {}
+
+
 class Monitoring(schema.Base):
     enabled: bool = True
+    overrides: MonitoringOverrides = MonitoringOverrides()
+    minio_enabled: bool = True
 
 
 class JupyterLabPioneer(schema.Base):
@@ -381,6 +389,12 @@ class DaskGatewayInputVars(schema.Base):
 
 class MonitoringInputVars(schema.Base):
     monitoring_enabled: bool = Field(alias="monitoring-enabled")
+    minio_enabled: bool = Field(alias="minio-enabled")
+    grafana_loki_overrides: List[str] = Field(alias="grafana-loki-overrides")
+    grafana_promtail_overrides: List[str] = Field(alias="grafana-promtail-overrides")
+    grafana_loki_minio_overrides: List[str] = Field(
+        alias="grafana-loki-minio-overrides"
+    )
 
 
 class TelemetryInputVars(schema.Base):
@@ -524,6 +538,14 @@ def input_vars(self, stage_outputs: Dict[str, Dict[str, Any]]):
 
         monitoring_vars = MonitoringInputVars(
             monitoring_enabled=self.config.monitoring.enabled,
+            minio_enabled=self.config.monitoring.minio_enabled,
+            grafana_loki_overrides=[json.dumps(self.config.monitoring.overrides.loki)],
+            grafana_promtail_overrides=[
+                json.dumps(self.config.monitoring.overrides.promtail)
+            ],
+            grafana_loki_minio_overrides=[
+                json.dumps(self.config.monitoring.overrides.minio)
+            ],
         )
 
         telemetry_vars = TelemetryInputVars(

src/_nebari/stages/kubernetes_services/template/modules/kubernetes/services/monitoring/loki/main.tf

@@ -0,0 +1,103 @@
+resource "random_password" "minio_root_password" {
+  length  = 32
+  special = false
+}
+
+locals {
+  minio-url = "http://${var.minio-release-name}:${var.minio-port}"
+  node-selector = {
+    "${var.node-group.key}" = "${var.node-group.value}"
+  }
+}
+
+resource "helm_release" "loki-minio" {
+  count      = var.minio-enabled ? 1 : 0
+  name       = var.minio-release-name
+  namespace  = var.namespace
+  repository = "https://raw.githubusercontent.com/bitnami/charts/defb094c658024e4aa8245622dab202874880cbc/bitnami"
+  chart      = "minio"
+  # last release that was Apache-2.0
+  version = var.minio-helm-chart-version
+
+  set {
+    name  = "accessKey.password"
+    value = "admin"
+  }
+
+  set {
+    name  = "secretKey.password"
+    value = random_password.minio_root_password.result
+  }
+
+  set {
+    name  = "defaultBuckets"
+    value = join(" ", var.buckets)
+  }
+
+  set {
+    name  = "persistence.size"
+    value = var.minio-storage
+  }
+
+  values = concat([
+    file("${path.module}/values_minio.yaml"),
+    jsonencode({
+      nodeSelector : local.node-selector
+    })
+  ], var.grafana-loki-minio-overrides)
+}
+
+
+resource "helm_release" "grafana-loki" {
+  name       = "nebari-loki"
+  namespace  = var.namespace
+  repository = "https://grafana.github.io/helm-charts"
+  chart      = "loki"
+  version    = var.loki-helm-chart-version
+
+  values = concat([
+    file("${path.module}/values_loki.yaml"),
+    jsonencode({
+      loki : {
+        storage : {
+          s3 : {
+            endpoint : local.minio-url,
+            accessKeyId : "admin"
+            secretAccessKey : random_password.minio_root_password.result,
+            s3ForcePathStyle : true
+          }
+        }
+      }
+      storageConfig : {
+        # We configure MinIO by using the AWS config because MinIO implements the S3 API
+        aws : {
+          s3 : local.minio-url
+          s3ForcePathStyle : true
+        }
+      }
+      write : { nodeSelector : local.node-selector }
+      read : { nodeSelector : local.node-selector }
+      backend : { nodeSelector : local.node-selector }
+      gateway : { nodeSelector : local.node-selector }
+    })
+  ], var.grafana-loki-overrides)
+
+  depends_on = [helm_release.loki-minio]
+}
+
+resource "helm_release" "grafana-promtail" {
+  # Promtail ships the contents of logs to Loki instance
+  name       = "nebari-promtail"
+  namespace  = var.namespace
+  repository = "https://grafana.github.io/helm-charts"
+  chart      = "promtail"
+  version    = var.promtail-helm-chart-version
+
+  values = concat([
+    file("${path.module}/values_promtail.yaml"),
+    jsonencode({
+    })
+  ], var.grafana-promtail-overrides)
+
+  depends_on = [helm_release.grafana-loki]
+}

src/_nebari/stages/kubernetes_services/template/modules/kubernetes/services/monitoring/loki/values_loki.yaml

@@ -0,0 +1,78 @@
+# https://github.com/grafana/loki/blob/4cae003ecedd474e4c15feab4ea2ef435afff83f/production/helm/loki/values.yaml
+
+loki:
+  storage:
+    type: s3
+  commonConfig:
+    replication_factor: 1
+  # Not required as it is inside cluster and not exposed to the public network
+  auth_enabled: false
+
+  # The Compactor deduplicates index entries and also apply granular retention.
+  compactor:
+    # is the directory where marked chunks and temporary tables will be saved.
+    working_directory: /var/loki/compactor/data/retention
+    # minio s3
+    shared_store: s3
+    # how often compaction will happen
+    compaction_interval: 1h
+    # should delete old logs after retention delete delay
+    # ideally we would want to do storage based retention, but this is not
+    # currently implemented in loki, that's why we're doing time based retention.
+    retention_enabled: true
+    # is the delay after which the Compactor will delete marked chunks.
+    retention_delete_delay: 1h
+    # specifies the maximum quantity of goroutine workers instantiated to delete chunks.
+    retention_delete_worker_count: 150
+
+  limits_config:
+    # The minimum retention period is 24h.
+    # This is reasonable in most cases, but if people would like to retain logs for longer
+    # then they can override this variable from nebari-config.yaml
+    retention_period: 60d
+
+  schema_config:
+    configs:
+      # list of period_configs
+      # The date of the first day that index buckets should be created.
+      - from: "2024-03-01"
+        index:
+            period: 24h
+            prefix: loki_index_
+        object_store: s3
+        schema: v11
+        store: boltdb-shipper
+  storage_config:
+    boltdb_shipper:
+        # Directory where ingesters would write index files which would then be
+        # uploaded by shipper to configured storage
+        active_index_directory: /var/loki/compactor/data/index
+        # Cache location for restoring index files from storage for queries
+        cache_location: /var/loki/compactor/data/boltdb-cache
+        # Shared store for keeping index files
+        shared_store: s3
+
+# Configuration for the write pod(s)
+write:
+  # -- Number of replicas for the write
+  # Keeping cost of running Nebari in mind
+  # We don't need so many replicas, if people need it
+  # they can always override from nebari-config.yaml
+  replicas: 1
+
+read:
+  # -- Number of replicas for the read
+  replicas: 1
+
+backend:
+  # -- Number of replicas for the backend
+  replicas: 1
+
+minio:
+  # We are deploying minio from bitnami chart separately
+  enabled: false
+
+monitoring:
+  selfMonitoring:
+    grafanaAgent:
+      installOperator: false

src/_nebari/stages/kubernetes_services/template/modules/kubernetes/services/monitoring/loki/values_minio.yaml

@@ -0,0 +1 @@
+# https://github.com/bitnami/charts/blob/440ec159c26e4ff0748b9e9866b345d98220c40a/bitnami/minio/values.yaml

src/_nebari/stages/kubernetes_services/template/modules/kubernetes/services/monitoring/loki/values_promtail.yaml

@@ -0,0 +1 @@
+# https://github.com/grafana/helm-charts/blob/3831194ba2abd2a0ca7a14ca00e578f8e9d2abc6/charts/promtail/values.yaml

src/_nebari/stages/kubernetes_services/template/modules/kubernetes/services/monitoring/loki/variables.tf

@@ -0,0 +1,84 @@
+variable "namespace" {
+  description = "deploy monitoring services on this namespace"
+  type        = string
+  default     = "dev"
+}
+
+variable "loki-helm-chart-version" {
+  description = "version to deploy for the loki helm chart"
+  type        = string
+  default     = "5.43.3"
+}
+
+variable "promtail-helm-chart-version" {
+  description = "version to deploy for the promtail helm chart"
+  type        = string
+  default     = "6.15.5"
+}
+
+variable "minio-helm-chart-version" {
+  description = "version to deploy for the minio helm chart"
+  type        = string
+  default     = "6.7.4"
+}
+
+variable "grafana-loki-overrides" {
+  description = "Grafana Loki helm chart overrides"
+  type        = list(string)
+  default     = []
+}
+
+variable "grafana-promtail-overrides" {
+  description = "Grafana Promtail helm chart overrides"
+  type        = list(string)
+  default     = []
+}
+
+variable "grafana-loki-minio-overrides" {
+  description = "Grafana Loki minio helm chart overrides"
+  type        = list(string)
+  default     = []
+}
+
+variable "minio-release-name" {
+  description = "Grafana Loki minio release name"
+  type        = string
+  default     = "nebari-loki-minio"
+}
+
+variable "minio-port" {
+  description = "Grafana Loki minio port"
+  type        = number
+  default     = 9000
+}
+
+variable "buckets" {
+  description = "Minio buckets"
+  type        = list(string)
+  default = [
+    "chunks",
+    "ruler",
+    "admin",
+    "loki"
+  ]
+}
+
+variable "minio-storage" {
+  description = "Minio storage"
+  type        = string
+  default     = "50Gi"
+}
+
+variable "minio-enabled" {
+  description = "Deploy minio along with loki or not"
+  type        = bool
+  default     = true
+}
+
+variable "node-group" {
+  description = "Node key value pair for bound resources"
+  type = object({
+    key   = string
+    value = string
+  })
+}

src/_nebari/stages/kubernetes_services/template/modules/kubernetes/services/monitoring/values.yaml

@@ -1 +1,7 @@
 # https://github.com/prometheus-community/helm-charts/blob/main/charts/kube-prometheus-stack/values.yaml
+
+grafana:
+  additionalDataSources:
+   - name: Loki
+     type: loki
+     url: http://loki-gateway.dev