High Availability (HA) Deployment

While OpenObserve can be installed and run in HA mode on bare metal servers, VMs, Kubernetes and possibly other platforms, we currently provide installation using helm charts officially. We may provide additional installers like terraform in future. Local disk storage is not supported in HA mode and as such configuring an object store is mandatory.

Helm charts

For accessing the object store we recommend the best security practice of using IAM roles wherever possible. In case of Amazon EKS you can use IAM roles for Service Accounts (IRSA).

You must download the values.yaml file and make required changes. At a minimum you must provide the details of bucket to be used for data storage and credentials (IAM role or keys) to access it. You can download the file using following command:

curl https://raw.githubusercontent.com/openobserve/openobserve-helm-chart/main/charts/openobserve/values.yaml -o values.yaml

Metadata store

You can use PostgreSQL or MySQL as metadata storage.

The official helm chart (starting Feb 23rd 2024) uses PostgreSQL as metadata store by default. It installs a PostgreSQL cluster (1 primary + 1 replica) for you. cloudnative-pg operator is used to install/manage the PostgreSQL cluster.

When using PostgreSQL or MySQL as metadata storage, etcd is still needed for cluster coordination.

If you use PostgreSQL or MySQL as metadata store, etcd will not store any data. If there are problems with etcd cluster, you can delete and reinstall the etcd cluster.

PostgreSQL

You don't need to do anything if you are using the official helm chart and want to use bundled PostgreSQL.

If you want to use external PostgreSQL, you can configure as below. You need to create the database first. In this example we use openobserve as database name.

config:
  ZO_META_STORE: "postgres"
  ZO_META_POSTGRES_DSN: "postgres://postgres:12345678@localhost:5432/openobserve"

postgres:
  enabled: false # disable bundled PostgreSQL

MySQL

You need create the database first, in the example we use openobserve as database name.

config:
  ZO_META_STORE: "mysql"
  ZO_META_MYSQL_DSN: "mysql://user:12345678@localhost:3306/openobserve"

Etcd

While using official helm chart, etcd for the cluster is already configured for you. You don't need additional configuration. if you want to use external etcd cluster, you can configure as below:

config:
  ZO_ETCD_PREFIX: "/zinc/observe/"
  ZO_ETCD_USER: ""
  ZO_ETCD_PASSWORD: ""
etcd:
  enabled: false # disable emebed etcd
  externalUrl: "my_custom_host.com:2379" # if bundled is false then this is required

Configuration

Amazon EKS + S3

You must set a minimum of 2 values:

S3 bucket where data will be stored

config:
  ZO_S3_BUCKET_NAME: "mysuperduperbucket"

IAM role for the serviceAccount to gain AWS IAM credentials to access s3

serviceAccount:
  annotations:
    eks.amazonaws.com/role-arn: arn:aws:iam::12345353456:role/zo-s3-eks

Once you have configured the above in your values.yaml file, you can run the below commands to install OpenObserve.

Follow AWS documentation to enable IRSA and create an IAM role that you can use. You can also refer to the IRSA introduction blog.

Sample IAM policy.

{
  "Id": "Policy1678319681097",
  "Version": "2012-10-17",
  "Statement": [
    {
      "Sid": "Stmt1678319677242",
      "Action": [
        "s3:PutObject",
        "s3:GetObject",
        "s3:ListBucket",
        "s3:DeleteObject"
      ],
      "Effect": "Allow",
      "Resource": "arn:aws:s3:::mysuperduperbucket/*"
    }
  ]
}