Building an S3-Compatible Object Storage Server on Bare Metal using MinIO

To build a highly available, distributed object storage cluster (which protects against drive and total server failures using Erasure Coding), MinIO requires a minimum of 4 drives. In this tutorial, we will use a robust 4-node bare-metal setup.

• minio-node1 (10.0.0.11)

• minio-node2 (10.0.0.12)

• minio-node3 (10.0.0.13)

• minio-node4 (10.0.0.14)

(Note: We assume you are running a fresh installation of Ubuntu 24.04 LTS on all nodes).

Execute this step on all four nodes.

First, ensure your servers are updated.

sudo apt update && sudo apt upgrade -y
                                

MinIO relies heavily on accurate hostname resolution to communicate across the internal network. Edit the /etc/hosts file on every server so they can locate each other:

sudo nano /etc/hosts
                                

Add the following lines to the bottom of the file on all four nodes:

10.0.0.11 minio-node1
10.0.0.12 minio-node2
10.0.0.13 minio-node3
10.0.0.14 minio-node4
                                

Execute this step on all four nodes.

MinIO strongly recommends using the XFS file system for the underlying storage drives due to its performance with massive files.

Assuming your dedicated server has a raw, unformatted NVMe drive located at /dev/nvme1n1, format it using XFS:

sudo mkfs.xfs /dev/nvme1n1 -L MINIO_DATA
                                

Next, create the mount point directory:

sudo mkdir -p /mnt/data1
                                

To ensure the drive mounts automatically if the server reboots, add it to your /etc/fstab file:

echo 'LABEL=MINIO_DATA /mnt/data1 xfs defaults,noatime 0 2' | sudo tee -a /etc/fstab
                                

Mount the drive immediately:

sudo mount -a
                                

Execute this step on all four nodes.

MinIO is distributed as a single, highly optimized binary file. Download the latest version directly from the official repository:

wget https://dl.min.io/server/minio/release/linux-amd64/minio
                                

Make the binary executable and move it to the system binary directory:

chmod +x minio
sudo mv minio /usr/local/bin/
                                

For security purposes, you should never run MinIO as the root user. Create a dedicated system user and group for the service:

sudo groupadd -r minio-user
sudo useradd -M -r -g minio-user minio-user
                                

Transfer the ownership of your mounted NVMe drive to this new user:

sudo chown minio-user:minio-user /mnt/data1
                                

Execute this step on all four nodes.

Create the MinIO configuration file directory and the file itself:

sudo mkdir /etc/minio
sudo nano /etc/default/minio
                                

Add the following configuration. This exact configuration must be identical across all four servers.

# Volume locations for the distributed cluster. 
# The {1...4} syntax tells MinIO to span across all 4 nodes.
MINIO_VOLUMES="http://minio-node{1...4}:9000/mnt/data1"

# Set the Root Access Key and Secret Key (These act as your AWS IAM Admin credentials)
MINIO_ROOT_USER="admin-minio-2026"
MINIO_ROOT_PASSWORD="SuperSecretStrongPassword!"

# The address to listen on for S3 API requests
MINIO_OPTS="--address :9000 --console-address :9001"
                                

(Save and exit the file).

Execute this step on all four nodes.

To ensure MinIO runs in the background and starts on boot, we will configure a systemd service.

Download the official MinIO systemd script:

cd /tmp
wget https://raw.githubusercontent.com/minio/minio-service/master/linux-systemd/minio.service
                                

Move it to the systemd directory:

sudo mv minio.service /etc/systemd/system/
                                

Reload the system daemon, start the MinIO service, and enable it on boot:

sudo systemctl daemon-reload
sudo systemctl start minio
sudo systemctl enable minio
                                

Verify that the cluster is healthy and running on all nodes:

sudo systemctl status minio
                                

(You should see active (running) and a log message indicating that the server has started in distributed mode).

Your massive, distributed object storage cluster is now online!

Open your web browser and navigate to the MinIO Console using any of your node IP addresses on port 9001: http://10.0.0.11:9001

Log in using the MINIO_ROOT_USER and MINIO_ROOT_PASSWORD you defined in Step 4.

From this beautiful Web GUI, you can:

• Click Buckets to create your first S3-compatible bucket (e.g., ai-training-data).

• Click Identity to generate Access Keys and Secret Keys for your developers.

• Monitor your cluster's NVMe drive health and network throughput.

To connect your existing applications to this cluster, simply update their AWS S3 SDK configurations to point to your new endpoint (http://10.0.0.11:9000) and replace the AWS credentials with your newly generated MinIO keys.

You have successfully built an enterprise-grade object storage cluster. Because of MinIO's erasure coding, you can safely lose an entire physical server without losing a single byte of your data.

As your data ingestion scales, avoiding network bottlenecks is critical. This is why enterprise architectures pair MinIO's highly parallel I/O capabilities with iDatam’s 100Gbps Dedicated Servers. By keeping your data on high-speed bare metal and routing it through unmetered pipes, you achieve faster-than-cloud performance while completely eliminating variable data egress costs.