HA Monolithic deployment

HA monolithic deployment is a replacement for the Simple Scalable Deployment (removed with Loki 4.0) where high availability is more important than scalability.

This example demonstrates how to run a number of single-binary Loki instances (-target=all) behind a reverse proxy (NGINX) for high availability and durability. The demo uses Docker compose, but the mechanics can be applied to any other deployment model, such as Helm or Tanka.

This folder contains a docker-compose.yaml and a config directory with configuration files for Loki and NGIX.

Usage

To use the latest features of Loki, you need to build the Docker image from main locally.

bash

make loki-image

Then replace the image in docker-compose.yaml.

bash

docker compose up

Architecture

mermaid

graph TB
    Client(["Client"])
    nginx["Load Balancer\n:3100"]
    subgraph ring ["Ring (RF=3)"]
        subgraph loki1_container ["loki-1"]
            loki1["-target=all\n:3100\n:9095\n:7946"]
            loki1 --- loki1_vol[("local disk")]
        end
        subgraph loki2_container ["loki-2"]
            loki2["-target=all\n:3100\n:9095\n:7946"]
            loki2 --- loki2_vol[("local disk")]
        end
        subgraph loki3_container ["loki-3"]
            loki3["-target=all\n:3100\n:9095\n:7946"]
            loki3 --- loki3_vol[("local disk")]
        end
    end


    objstore[("Object Storage\n(S3, GCS, Azure Blob, ...)")]

    Client -->|"HTTP :3100"| nginx

    nginx ---> loki1
    nginx ---> loki2
    nginx ---> loki3

    loki1 & loki2 & loki3 -----|"write/read"| objstore