Observability: Sentry
This course covers error tracking with Sentry in a Rails application. It explains how an error tracker groups identical exceptions into counted issues, and how that differs from a log aggregator. It walks through installing sentry-ruby and sentry-rails, setting environment and release, sampling, and privacy options such as before_send. It then covers fingerprints and noise, user context, tags, breadcrumbs and job scopes, triage by users affected, and incident response with severity levels and blameless postmortems. It ends with a full configuration for one application.
What you'll learn
- ✓How grouped exceptions answer questions a stream of log events cannot
- ✓Installing and configuring Sentry, including release, sampling and privacy options
- ✓Fixing over-grouping and under-grouping with bounded fingerprints, and removing noise at the source
- ✓Adding user context, tags, breadcrumbs and job scopes to make errors reproducible
- ✓Triaging issues by users affected and handling regressions
- ✓Incident response with severity levels and blameless postmortems
An exception is not a log line
Your application raised a NoMethodError in production. What happens next depends entirely on which tool receives it.
Read the full lesson free →Curriculum
7 lessonsAn exception is not a log line
How an error tracker groups, counts and flags exceptions, and what context arrives with each one.
Installing and configuring Sentry
Installing sentry-ruby and sentry-rails, choosing environment, release, sampling and privacy options, then sending a test event.
Grouping, fingerprints and noise
How fingerprints group events, how over-grouping and under-grouping happen, and where to remove noise.
Context: making an error reproducible
Setting user ids, bounded tags, extra context, breadcrumbs and job scopes, and linking issues to traces.
Triage: what to do with a thousand issues
Ranking issues by users affected, using issue states, handling regressions, clearing a backlog and assigning owners.
Incident response
A response sequence, severity levels defined in advance, the four tools in order, and blameless postmortems.
Putting it together: error tracking for this app
The full configuration for this application, a triage cadence, and the ideas that carry across tools.