CedarDB as Server

CedarDB as Server

This guide will show you how you can launch CedarDB as a server, create users and configure it to accept outside connections.


CedarDB currently supports Linux. You can run it standalone without any system setup dependencies.

We have extensively tested CedarDB on Ubuntu, Debian, and Arch Linux. As runtime environment, CedarDB only needs a glibc as shipped by Debian oldstable (glibc 2.31, released early 2020) or newer.

The CedarDB distribution comes with a binary called server, located in the bin directory. This binary launches a server daemon that allows outside connections to the database.

As with any server process, you should be mindful how this process can interact with the rest of the system and which users can use it.

Launching the server

Create a database

The server needs to connect to an existing database. If you have followed the Quick Start Guide you did already create a database and can skip this step.

Otherwise, create a new database directory:

mkdir mydb
CedarDB is optimized for modern storage hardware. To ensure you get CedarDB’s best performance, place this directory on a reasonably modern SSD.

Then start the sql shell, instructing it to initialize a new database (called test, for example):

bin/sql --createdb mydb/test.db

finally exit the shell with CTRL-D.

Launch the server process

Run the server interactively:

bin/server mydb/test.db

After this initial startup, the CedarDB server does not allow arbitrary connections. During this state, you can only connect via a local domain socket as the same OS user that also launched the server process.

Connect to the server via the psql shell

The server speaks the PostgreSQL wire protocol making it compatible with a wide array of PostgreSQL supporting tools, the most important being psql, the default terminal-based front-end to PostgreSQL.

All major Linux distributions provide packages containing psql. On Ubuntu, for example, it is provided by the postgresql-client package.

Connect to the server in a separate terminal via a local domain socket as the postgres DB super user:

psql -h /tmp -U postgres

Create a new user

You probably do not want to execute all your queries as the default superuser, or only connect via domain socket, so create a new user and initialize it from within the psql shell:

-- Create a user that can manage the database
create user test superuser; -- You can also grant rights more fine-grained.
create database test; -- By default the user will use the database with its name, so let's create one.
alter user test with password '1234';

Connecting to the server

You can now connect to CedarDB with your newly created user over IP:

psql -h localhost -U test

If you want to connect from a different machine, you have to instruct CedarDB to accept non-local connections:

# '::' specifies all IPv4 and IPv6 connections
bin/server mydb/test.db --address=::

Afterward, you can connect remotely:

psql -h CedarDB.example.com -U test

Manage CedarDB with systemd

You can optionally let systemd automatically launch the CedarDB server whenever a new connection request comes in using its Socket Activation feature.

First create the following service and socket file, replacing the path to the CedarDB server binary and database file with your own values.

Description=CedarDB server

ExecStart=path/to/bin/server -verbose path/to/your/db/dbname.db
Description=Socket activation unit for CedarDB server


You can now use the following script to create an CedarDB service and register it with systemd:

#!/usr/bin/env bash
set -euo pipefail

install <(< cedardb-server.service sed "s:XXX:$PWD:g") "$SYSTEMD_PATH/cedardb-server.service"
install cedardb-server.socket "$SYSTEMD_PATH/cedardb-server.socket"
systemctl --user daemon-reload

You have now installed the CedarDB systemd service. Start socket activation with:

systemctl --user start cedardb-server.socket"