TimescaleDB is an extension of Postgres that is required for the Flags feature. It is recommended to use the Virtual Machine/EC2 installation for TimescaleDB instead of using Helm.

We have included two different ways for the Timescale implementation:

Primary/Secondary servers with replication
Single Server implementation.

We recommend using Primary/Secondary for production-level environments and the Single Server for non-production environments. The TimescaleDB implementation is stored in the TimescaleDB folder, such as https://github.com/codecov/enterprise-resources/tree/master/google_kubernetes_engine/timescaledb.

Primary/Secondary Servers

We have created a template that will create two virtual machines and will run everything required for TimescaleDB, except for setting the password for the Postgres user. If you want to enable Primary/Secondary Servers, add a value of “true” to the config.auto.tfvars for timescale_server_replication_enabled config value. This creates 2 VMs that will replicate from a primary server to a secondary server. Also, we have configured pgBackRest to perform full and incremental backups to a storage account. You will need to update the timeseries_database_url config value and set it to the primary server (example: postgres://username:pw@primaryserver:5432) and update the timeseries_enabled config value to “true” in the codecov.yml file to allow enable the feature.

Single Server

We have created a template that will create a virtual machine and will run everything required for TimescaleDB, except for setting the password for the Postgres user. If you want to enable Single Server Implementation, add a value of “false” to the config.auto.tfvars for timescale_server_replication_enabled config value. Also, we have configured pgBackRest to perform full and incremental backups to a storage account. You will need to update the timeseries_database_url config value (example: postgres://username:pw@server:5432) and update the timeseries_enabled config value to “true” in the codecov.yml file to enable the feature.

Managed Service

TimescaleDB also offers a managed service. This can be used if you wish to abstract management burdens.

Helm and Kubernetes

TimescaleDB offers a helm chart for usage with Kubernetes. This method of deployment is not supported by the Codecov team but can be used to deploy TimescaleDB in your cluster.