Tamland - Infrastructure Capacity Forecaster¶

Tamland forecasts capacity for infrastructure components and predicts saturation events. Its intention is to prevent infrastructure-related incidents and provide a less obstrusive and earlier warning mechanism compared to sending out alerts when saturation occurs.

Tamland is implemented in Python and consumes Prometheus saturation metrics and uses Facebook Prophet to forecast future saturation.

GitLab uses Tamland to forecast infrastructure capacity and assist with capacity planning purposes. For GitLab specific documentation on how to use Tamland, please refer to the available documentation in the GitLab Handbook.

Installation¶

Tamland can be used from a development environment or installed through Docker (see below). A Helm chart is also available for Kubernetes.

Docker¶

# Latest release:
registry.gitlab.com/gitlab-com/gl-infra/tamland:latest

# Versioned:
registry.gitlab.com/gitlab-com/gl-infra/tamland:4.0.0

Usage: Tamland Shell¶

Tamland comes with a CLI tool shell, which can be invoked through executing bin/shell.

A help command is available, which gives an overview over available commands. Similarly, help is available for any of the commands implemented, e.g. invoke bin/shell forecast --help to get usage information for the forecast command.

bin/shell --help
requirements.txt updated, running pip install

 Usage: shell.py [OPTIONS] COMMAND [ARGS]...

[...]

Please refer to the forecasting page to get started with forecasting.

Environment variables¶

The following table contains commonly used environment variables.

variable	description		default
LOGLEVEL	Log-level to be used for file based logging (tamland.log)	optional	INFO

Trivia and History¶

Brick Tamland, weather forecaster

The project name was based on a character Brick Tamland, the weather forecaster from the movie Anchorman.