Skip to main content

Installation

Analytics Data Processing and Analytics API can be installed with Docker or as Windows services. Both are supported. We recommend Docker: you upgrade or roll back by changing a version number, and the images include all dependencies, such as the database driver.

The installation pages apply to all current versions. What differs between versions is in the release notes and the configuration reference for each version, see Settings.

Before you start​

You need:

  • A server with access to both databases below. Linux or Windows for Docker, Windows for the Windows installation.
  • The EWA database (Microsoft SQL Server). Analytics Data Processing only reads from it.
  • An empty database for Analytics (Microsoft SQL Server). Analytics Data Processing creates and fills the tables, and Analytics API reads from them.
  • Access to the Docker images or the Windows packages. Contact support to get access.

Check Version Compatibility to find the versions that match your EWA version.

Settings​

Both installations use the same settings, set in a .env file per service. All settings are described in the configuration reference for each version, for example Analytics Data Processing v2.6.0 and Analytics API v1.4.13.

The services read their settings only when they start. After changing a .env file, restart the service, see Restarting.

First run​

Every time Analytics Data Processing starts, it runs twice in a row: first a start-up run, then the first scheduled run. After that, it runs on the schedule, by default every 60 minutes. Each run ends with Transformer completed! in the log. The first run is finished when the log shows it twice.

On a new installation, the start-up run creates and fills the tables in the Analytics database. This can take a while for a large EWA database.

Analytics API reads the table structure when it starts. Restart it when the first run is finished, so it sees the new tables. Then verify it, see How to use the API.

Full load and transform​

Some upgrades rebuild tables with a full load, a full transform, or both. The Notes in the release notes say which, and this is how you do it:

  1. Before you start the new version, set BLIKSUND_LOAD_FULL_ON_START=true and other settings as the release notes say.
  2. Start the service.
  3. Set the values back to false. The running service keeps the values it started with, so the full load or transform still runs.
  4. When the first run is finished, restart both services: Analytics Data Processing so it uses false, and Analytics API so it sees the rebuilt tables.

If the release notes does not say to use full load or transform at start, starting the service once and letting it run is fine.

Upgrading​

Upgrades are never automatic. You choose when to upgrade:

  1. Find the new versions in Version Compatibility, and read the release notes for every version between your current version and the new one.
  2. Update the .env files if the Notes ask for new or changed settings.
  3. If the Notes ask for a full load or transform, follow the steps above. Otherwise, install the new version, and restart Analytics API when the first run is finished.

How you install a new version is described in Docker Installation and Windows Installation.

Restarting​

Restart a service after changing its .env file, and restart Analytics API after Analytics Data Processing has created or changed tables. The commands are in Docker Installation and Windows Installation.