PyPGIO - An IO generator for PostgreSQL based on the original pgio and SLOB

PyPGIO is a Python based I/O generator for PostgreSQL databases. The only purpose is to drive (a lot of) I/O from a Postgres database without requiring a lot of CPU resources.

It is based on, and uses some code from PGIO by Kevin Closson, see PGIO for more details.

• Linux shell account (can be different system than the database server)
• $HOME/bin directory that is in $PATH
• PostgreSQL database (any recent supported version)
• User on the database with full access rights on the database
• Python 3.6 or higher
• python3 must be in the environment
• Internet access to install Python PIP packages

Assuming the prerequisites are met, the easiest way to install the latest version of PyPGIO is to run the downloader command on your host:

# Test if python3 works
python3 --version

# This requires internet access via https
# Inspect downloader (optional)
curl https://raw.githubusercontent.com/bsjerps/pypgio/master/scripts/download | less

# Download using downloader
curl https://raw.githubusercontent.com/bsjerps/pypgio/master/scripts/download | bash

# Check if pypgio is installed
ls -al ~/bin/pypgio

Note that PyPGIO is designed to run as a ZipApp package. Running directly from the git repository is not recommended. Instead, create the package from the repo:

# Clone the repository
git clone https://github.com/bsjerps/pypgio.git
cd pypgio
scripts/mkapp
ls -al ~/bin/pypgio

PyPGIO depends on the Python modules psycopg and PrettyTable. These modules are not installed by default with Python. In order for PyPGIO to work, it needs to run from a Python virtual environment in which these modules are installed.

To make life easier if you are not familiar with Python, PyPGIO has a built-in installer module to setup the environment automatically. The virtual environment will be in $HOME/pgio_venv

# Show the installer help
python3 $HOME/bin/pypgio/install -h

# Setup virtualenv, pgio wrapper script and bash completions
python3 $HOME/bin/pypgio/install -h

# Uninstall virtualenv, wrapper script and completions file
python3 $HOME/bin/pypgio/install -u

# Logout and login again - bash completion should now work
# Or source the completions
source ~/.bash_completion

See create_db.sql for details

# Run pgio
pgio

# Show help summary: Run pgio -h
pgio -h

# Show default configuration settings
pgio configure

# Show configuration parameters
pgio configure -h

# Change the database host
pgio configure --dbhost pgserver.lan

# Test connection
pgio

Assuming we have a database named pgio with a user pgio and password pgio. By default, pgio uses the public table with the default schema.

On RHEL, the database top directory is /var/lib/pgsql/<version>/data

But in many cases, we want the (large) pgio tables on another filesystem. In our example we have a file system /pgdata on which we created a PostgreSQL tablespace bulk and we want our tables to be generated there.

# Set schemas and scale
pgio configure --schemas 8 --scale 256M

# Optionally set the default tablespace
pgio configure --tablespace bulk

# Create database structure and tables, using 4 parallel threads
pgio setup 4

# Check the tables
pgio list

Now we can run pgio:

# Run pgio for 60 seconds with 8 threads
pgio run 60 8

# Change the update percentage
pgio configure --update_pct 10

# Run pgio for 60 seconds with 4 threads
pgio run 60 4


# View detailed reports
pgio report -v

Deinstallation involves the removal of all pypgio related stuff in the database, and the virtual environment and shell settings in the user home directory

# Delete the database structures
pgio destroy

# Remove bash setup and virtual environment
python3 $HOME/bin/pypgio/install -u

# Remove pypgio from $HOME/bin
rm ~/bin/pypgio