Kubernetes logging by Loki stack (Loki+Promtail)

Terraform module for deploy Loki logging to your kubernetes cluster, with multi cloud storage support.

Wokrflow

Module creates all necessary resources for logging important containers inside your kubernetes cluster. Previously you need to have Grafana to see your logs. Loki is only separate Data Source for Grafana. Module supports different storages for logs: AWS S3 bucket, GCP GCS Bucket, Azure Blob storage and Kubernetes Persistent Volume.

Software Requirements

Name	Description
Terraform	>= v0.14.9
Helm provider	>= 2.1.0
Kubernetes provider	>= v2.0.1

Usage

AWS with S3 as storage

module "aws_s3_loki_stack" {
  source  = "terraform-iaac/loki-stack/kubernetes"

  # In case if IRSA is enabled. IRSA must have S3 RW Policy access.
  # Otherwise, your instance must have S3 RW Policy attached.
  loki_service_account_annotations = {
    "eks.amazonaws.com/role-arn" = "arn:aws:iam::123456789:role/loki-logging"
  }

  provider_type = "aws"
  s3_name       = "s3-bucket-loki-logs"
  s3_region     = "us-east-1"
}

GCP with GCS as storage

module "gcs_loki_stack" {
  source  = "terraform-iaac/loki-stack/kubernetes"
  
  # In case if Workload Identity is enabled.
  # Otherwise, your node must have RW permissions to GCS.
  loki_service_account_annotations = {
    "iam.gke.io/gcp-service-account" = "[email protected]"
  }

  provider_type   = "gcp"
  gcs_bucket_name = "gcs-bucket-loki-logs"
}

Azure with Blob as storage

module "azure_loki_stack" {
  source  = "terraform-iaac/loki-stack/kubernetes"

  provider_type              = "azure"
  storage_account_name       = "kuberneteslogging"
  storage_account_access_key = "super-secret-key"
  container_name             = "logs"
}

PV local as storage

module "pv_local_loki_stack" {
  source = "terraform-iaac/loki-stack/kubernetes"

  provider_type          = "local"
  pvc_storage_class_name = "default"
  pvc_access_modes       = ["ReadWriteOnce"]
  persistent_volume_name = kubernetes_persistent_volume.pv_loki.metadata.0.name
  persistent_volume_size = "4Gi"
}

Note: provider_type supports only aws, azure, gcp or local value. Every value require own variables (see locals section in varaibles.tf file or check examples.)

Inputs

Name	Description	Type	Default	Example	Required
namespace	Name of namespace where you want to deploy loki-stack	string	monitoring	n/a	no
create_namespace	Create namespace by module? true or false	bool	true	n/a	no
loki_resources	Compute Resources required by loki container. CPU/RAM requests	map	{ request_cpu = "50m" request_memory = "100Mi" }	{ request_cpu = "20m" request_memory = "50Mi" }	no
promtail_resources	Compute Resources required by promtail container. CPU/RAM requests	map	{ request_cpu = "20m" request_memory = "50Mi" }	{ request_cpu = "20m" request_memory = "50Mi" }	no

Loki variables

Name	Description	Type	Default	Example	Required
loki_name	Loki application name	string	loki	n/a	no
loki_docker_image	Image for Loki container	string	grafana/loki:2.3.0	n/a	no
loki_termination_grace_period_seconds	Grace period applies to the total time it takes for both the PreStop hook to execute and for the Container to stop normally	integer	4800	n/a	no
loki_port	Port mapping to kubernetes service	list(object({ name = string internal_port = integer external_port = integer }))	[ { name = "http-metrics" internal_port = 3100 external_port = 3100 } ]	n/a	no
loki_node_selector	Select node to deploy loki stack	map	null	{ (local.node_spot_label_key) = false (local.node_multi_az_label_key) = true }	no
loki_toleration	Loki Pod node tolerations	list(object({ effect = string // (Optional) key = string // (Optional) operator = string // (Optional) toleration_seconds = number // (Optional) value = string // (Optional) }))	[]	[ { effect = "NoSchedule" key = "gpu" operator = "Equal" value = "true" } ]	no
loki_service_account_annotations	Add additional account annotations to Loki service account	map	ReadWriteMany	n/a	no

Promtail

Name	Description	Type	Default	Example	Required
promtail_name	Promtail application name	string	monitoring-alertmanager-pv	n/a	no
promtail_docker_image	Image for Promtail container	string	2Gi	n/a	no
promtail_internal_port	Port mapping to daemon-set	list(object({ name = string internal_port = integer }))	[ { name = "http-metrics" internal_port = 3100 } ]	n/a	no
promtail_toleration	Promtail pod node tolerations	list(object({ effect = string // (Optional) key = string // (Optional) operator = string // (Optional) toleration_seconds = number // (Optional) value = string // (Optional) }))	[]	[ { effect = "NoSchedule" key = "gpu" operator = "Equal" value = "true" } ]	no

Storage variables

Name	Description	Type	Default	Example	Required
provider_type	Choose what type of provider you want (aws, azure, gcp and local)	string	n/a	azure	yes

AWS S3

Name	Description	Type	Default	Example	Required
s3_region	AWS region where s3 locate	string	null	us-east-1	no
s3_name	Name of s3 bucket	string	null	s3-bucket-logs	no

GCP storage

Name	Description	Type	Default	Example	Required
gcs_bucket_name	Google Cloud Storage bucket name	string	null	gcs-storage-logs	no

Azure Blob Storage

Name	Description	Type	Default	Example	Required
storage_account_name	The Microsoft Azure storage account name to be used	string	null	aks-application-logs	no
storage_account_access_key	The Microsoft Azure storage account access key to use	string	null	sEcRetKeY	no
container_name	Name of the blob container used to store chunks. This container must be created before running cortex.	string	null	my-app-logs	no

Local

Name	Description	Type	Default	Example	Required
persistent_volume_name	Name of persistant volume	string	null	k8s-my-app-logs-pv	no
persistent_volume_size	Name of persistant disk size	string	null	4Gi	no
pvc_access_modes	Mode for access to data	string	null	["ReadWriteOnce"]	no
pvc_storage_class_name	Type of storage class name	string	null	default	no