Send errors, traces and metrics
Instrument your application so problems point at the release that caused them.
1Get a key
Monitoring → Setup → Ingest keys → Create key, for the environment. The checklist on the same page tracks your first event, monitor, alert destination and release.
2Server logs
On a server’s Logs tab, under Send logs to Monitoring, choose an environment. A small agent then sends the server’s system warnings and errors, and its websites’ Laravel warnings and errors with stack traces, to Monitoring → Events (type log), where you can search them and set alert rules on matching lines. It sends up to 600 lines a minute and uses its own key, revoked when you stop.
3Tickets
Monitoring → Setup → Ticket trackers: connect GitHub Issues (a repository and a fine-grained token that can write issues), Linear (an API key and the team ID) or Jira Cloud (your atlassian.net site, email, API token and project key). Then choose Create ticket on any issue; the ticket gets the error, where it happened, how often, and a link back, and the issue links to it. When the ticket is closed (completed in Linear, done in Jira), the issue is resolved within the hour; if the error happens again, it reopens.
4Laravel
For Laravel apps, composer require buildpusher/laravel and set BUILDPUSHER_TOKEN. Each request is recorded with a timeline of its queries, mail and notifications and its cache hits and misses; exceptions carry a hashed user ID so issues count the people affected; slow queries and N+1 queries (the same query five or more times in a request) are flagged; queue jobs, scheduled tasks, commands and warning-level logs are recorded too. php artisan buildpusher:deploy marks each release.
5Send data
Use the SDK snippet on the setup page, or send OpenTelemetry traces, logs and metrics to the OTLP endpoint shown there.
6Investigate
Errors are grouped into Issues; Events, Metrics and Traces let you dig in, and Releases show what changed.
7Browser errors
Monitoring → Setup → Browser errors. List the origins your pages are served from and turn it on, then add the script tag it shows to every page. JavaScript errors and unhandled promise rejections become issues from the “browser” service, with the browser, operating system and release. Query strings are dropped from page addresses.
8Link to deploys
Every trace shows the deploy that was live when it ran, and each deploy links to its requests and traces, so you can see what a release changed.