Attribute distributions

Attribute distributions can help you get to the source of an error by showing a distribution of errors per tag.

Screenshot of Attributes Page

This can help you show how many errors happened on frontend1 vs frontend2, for example. Or if a specific deploy caused more errors than another deploy.

Attribute distributions work by looking at "special" tags in a sample. We extract the values from these special tags and compute the distributions from them.

Tagging samples

You can tag a sample with the following tags to get distributions for them:

  • hostname To show distributions by hosts
  • revision To show distributions per deploy/revision
  • country (For front-end errors) to show which countries experience the most issues
  • browser (For front-end errors) to show which browser is the cause
  • user_id To see which users are impacted
  • company If your app has a company level above users

Most of these tags are applied to samples automatically; our agent sets the hostname tag and if you use deploy tracking with ENV vars, we use the revision tag from that.

For front-end errors we extract the country from the IP address that sent the error, as well as the browser tag.

Tags we don't set automatically are the user_id and company tags. You can tag these yourself to get a better sense of which users have experienced an issue.

Notes to keep in mind

We extract tags from samples, which means in some cases, if an enormous amount of errors are sent in a short timeframe, we don't get every tag from every sample, as these are throttled by our integrations.

Tags are stored for the duration of your plan's retention. This can range from seven days to sixty days, depending on your plan. This means distributions can change over time, depending on how many error samples we still have in retention.