Three machines with systemd

A production TeselaDB cluster is three or more machines with the same binary, and here you set it up on three Linux machines, each node as a systemd service. At the end you reboot one of them to see the cluster keep serving.

What you will build

Three nodes, node-01, node-02 and node-03, joined with enrolment tokens and started by systemd every time their machine boots, with your applications connecting to the three of them with TLS and the user you create.

Requirements

• Three Linux machines with the binary at /usr/local/bin/teseladb (Install on Linux) and clocks synchronised with NTP.
• Port 27100 open between the three machines, 27017 towards your applications and 27200 towards your operations team.

1. The service on the three machines

Create /etc/systemd/system/teseladb.service on each machine. The unit starts the node once the network is up and restarts it if it fails:

[Unit]
Description=TeselaDB
After=network-online.target

[Service]
ExecStart=/usr/local/bin/teseladb start --dbpath /var/lib/teseladb --bind_ip 0.0.0.0 --peer-listen 0.0.0.0:27100 --admin-listen 0.0.0.0:27200
Restart=on-failure

[Install]
WantedBy=multi-user.target

2. Start the first node

On node-01, the first start creates the cluster CA and the root user:

sudo systemctl enable --now teseladb

3. Join the other two

On node-01, create a single-use token for each of the other two nodes. Each token expires after one hour, as --ttl 1h sets, so create them right before you join node-02 and node-03 to the running cluster:

teseladb admin --dbpath /var/lib/teseladb token create --ttl 1h

On node-02 and on node-03, the first start carries --join and the token. Then stop that process. Hand the node to the service, which no longer needs the token:

teseladb start --dbpath /var/lib/teseladb --bind_ip 0.0.0.0 --peer-listen 0.0.0.0:27100 \
  --admin-listen 0.0.0.0:27200 --join node-01:27100 --enroll-token <token>
sudo systemctl enable --now teseladb

4. Connect your applications

Copy the cluster CA, /var/lib/teseladb/pki/ca.pem, to your application machines and create a user for each one with the roles it needs (Users and roles). The string carries the three nodes:

mongodb://shop-api:<password>@node-01:27017,node-02:27017,node-03:27017/shop?tls=true&tlsCAFile=./ca.pem

Try it

teseladb admin --dbpath /var/lib/teseladb nodes lists the three nodes as active. Reboot node-03 with sudo reboot while your application writes: the ranges it led elect a new leader on the other two nodes and the driver retries the operation that was in progress. When the machine comes back, systemd starts the node and it catches up on its own.

Clean up

If it was a test, stop the service and delete the data on each machine, with the CA and the keys of the cluster:

sudo systemctl disable --now teseladb
sudo rm -r /var/lib/teseladb

Next steps

• Production checklist, before taking real traffic.
• Scheduled backups and a restore test, for the first nightly backup.