Stream Supabase Postgres changes into RawTree with logical replication and the supabase/etl worker.

Supabase Postgres CDC

This example streams a Supabase Postgres table into RawTree with Change Data Capture.

Superstore Sales Analytics via Supabase CDC It imports the Kaggle Superstore Sales dataset into Supabase, publishes the table through logical replication, and runs a supabase/etl worker that lands the initial copy plus inserts, updates, and deletes in an append-only RawTree table.

The walkthrough uses the Kaggle Superstore Sales dataset as sample data. Use the Showroom version for the curated dashboard and live walkthrough.

What this demonstrates

Initial copy plus live CDC from Supabase Postgres.
Append-only event storage in RawTree.
Terraform-managed deployment to ECS Fargate.
A build-from-source path for understanding and debugging the worker.
Dashboard queries that reconstruct the latest table state from CDC events.

1. Prepare the source table in Supabase

Create a Supabase project and import the Superstore Sales CSV through Table Editor → New table → Import data from CSV. Name the table superstore_sales_data in the public schema.

Supabase preserves the CSV column names verbatim, including spaces such as Row ID and Order ID.

Enable logical replication for the table:

-- REPLICA IDENTITY FULL makes UPDATEs and DELETEs include the full old row,
-- which the worker needs to emit complete CDC events.
ALTER TABLE public.superstore_sales_data REPLICA IDENTITY FULL;

DROP PUBLICATION IF EXISTS rawtree_superstore_publication;
CREATE PUBLICATION rawtree_superstore_publication
FOR TABLE public.superstore_sales_data;

2. Get the connection URL and CA certificate

Copy the Direct connection string from Project Settings → Database → Connection string. Do not use the pooler URL; logical replication requires a direct replication connection.

Download the Supabase CA certificate from Project Settings → Database → SSL Configuration.

You can also dump the CA from the live TLS handshake:

host=db.<your-project-ref>.supabase.co
openssl s_client -showcerts -starttls postgres -connect "$host:5432" </dev/null 2>/dev/null \
  | awk '/-----BEGIN CERT/,/-----END CERT/' > ~/supabase-ca.crt

Supabase direct endpoints are commonly IPv6-only. If you deploy to ECS, use a dual-stack subnet so the worker can reach Supabase.

3. Configure credentials

Set RawTree credentials as environment variables. The AWS provider uses your default AWS credentials or AWS_PROFILE.

export RAWTREE_API_KEY="rt_..."
export RAWTREE_ORG="your-org"
export RAWTREE_PROJECT="your-project"

export TF_VAR_supabase_database_url='postgres://postgres:PASS@db.<ref>.supabase.co:5432/postgres?sslmode=require'

supabase/etl

supabase/etl is Supabase's Rust framework for Postgres logical-replication pipelines. It handles the initial copy, streams ongoing inserts, updates, and deletes, and lets destinations write those events wherever they need to go. See the supabase/etl documentation for the upstream project details.

4. Deploy with Terraform

A single rawtree_supabase_cdc_ingestion resource provisions the worker infrastructure: Secrets Manager, IAM, CloudWatch Logs, ECS cluster, and a long-running Fargate service.

terraform {
  required_providers {
    rawtree = { source = "rawtreedb/rawtree" }
    aws     = { source = "hashicorp/aws", version = "~> 5.0" }
  }
}

provider "rawtree" {}
provider "aws" { region = "us-east-1" }

variable "supabase_database_url" {
  type      = string
  sensitive = true
}

resource "rawtree_supabase_cdc_ingestion" "superstore" {
  name        = "superstore"
  region      = "us-east-1"
  publication = "rawtree_superstore_publication"

  # For production, prefer database_url_secret_arn and
  # tls_root_cert_secret_arn pointing at secrets you manage.
  # Those values never enter Terraform state.
  database_url      = var.supabase_database_url
  tls_root_cert_pem = file("~/supabase-ca.crt")

  subnet_ids       = [aws_subnet.this.id] # dual-stack subnet
  assign_public_ip = true                 # IPv4 for the ghcr.io image pull

  cpu    = 512
  memory = 1024
}

output "log_group_name" {
  value = rawtree_supabase_cdc_ingestion.superstore.log_group_name
}

5. Apply and verify

Apply the configuration and tail the worker logs. On first start, the worker copies the existing Superstore rows into public_superstore__sales__data, then streams live changes from the replication slot.

terraform init
terraform apply

aws logs tail $(terraform output -raw log_group_name) --follow

rtree query "SELECT count() FROM public_superstore__sales__data"

6. Open the dashboard

Create a read-only API key for the dashboard:

rtree key create --name superstore-dashboard --permission read

Then open the Showroom dashboard. Insert, update, or delete rows in Supabase and watch the CDC operations update within seconds.

Supabase Postgres CDC

On this page