Configure access logging : Referrers

[jvwong]

Background

It would be desirable to capture where visitors to the site are coming from and their respective frequency. Detail about these referrers can help us maintain and inform new ways to reach out to users at journals and elsewhere.

To date, this information has been gleaned from a Google Analytics (GA) script attached to the homepage, but there are many limitations for this tool (e.g. Ad blockers). Also, this information can be precluded from requests by setting HTTP headers Referrer-Policy or at the HTML tag level.

Goal

To address the limitations of GA blocking, collect as much useful information via access logs. Currently, there are two opportunities to configure logging:

• Reverse proxy
  • An Nginx instance
• Web application
  • Node/ExpressJS HTTP server

Details

• Currently
  • Nginx access logs are defaulted to redirect to stdout.
  • Node/Express collects minimal access information via winston, which redirects to stdout
    • I don't believe its configured to print out referrer information
• Files
  • Log rotation. Another issue to keep in mind is that Docker containers make any associated data ephemeral, unless e.g. mapped to a volume.