Loki High Availability Mechanism


I deployed Loki using Ec2 instance. Which is used as reader as well as writer. Now I want two more Ec2 instances to implement high availability. I am aggregating logs of my ECS services using Loki Primary instance.

  • Is it possible to configure High Availability mechanism using other instances.

  • How we can specify master node and backup node?

  • Is there any way to configure if master node fails then secondary nodes (Instance) will act as a master and once a master is up again it will take over aggregating logs and pushing to S3? As I am storing logs in Loki server as as well backing up in S3 bucket.

Hi @ownrover13

I believe there are several scenarios documented in Deployment modes | Grafana Loki documentation. I hope this helps.


1 Like

Hi @eyeveebee

Thank you for the reply.

I already checked with the documentation and deployed the environment accordingly only. The thing is I can read and write the logs from grafana and to Loki respectively. But if my loki server goes down then I want to implement DR mechanism so that my log pulling and pushing will not get disturbed.

Thank you!

Assuming your configuration is correct (mostly the ring membership part that’s related to clustering), you can scale Loki by simply giving it more nodes. There is no master or slave in Loki, all nodes take traffic. You’ll of course also need some sort of load balancer as frontend.