Skip to content
← Telemetry Insights for Laravel 1.0.1 View on GitHub

Telemetry Insights

Telemetry Insights

cboxdk/laravel-telemetry-insights adds the parts of observability that need to remember something: an issue ledger, correlated incidents, alert rules and periodic digests.

It is an add-on to cboxdk/laravel-telemetry-ui. The dashboard stays what it is — a read-only window onto Tempo, Loki and Prometheus. This package reads through the same query layer and keeps its own small set of tables beside it.

Issues: one row per error fingerprint, with the status the team gave it — open, snoozed, resolved in a named release — rather than one row per occurrence.

Incidents: error groups that moved together, with the suspected cause and the evidence for it — a dependency that started failing, or a deploy the onset followed.

Both screenshots are captured by this package's own browser tests, against seeded fixture data, so what is pictured is what the code renders. See src/Testing/ScreenshotManifest.php.

Why a separate package

The dashboard queries your telemetry stores and writes nothing to them. That is a checkable property, not a slogan, and it is why you can point a desktop client at production.

Some things cannot be answered that way. "Has this exception ever happened before?" and "did this come back after we fixed it?" need a ledger. A cooldown needs to know when you were last paged. So those live here, in tables of our own — never in your telemetry stores, which stay untouched.

Install this and you get state. Do not, and the dashboard is exactly as read-only as it was.

What it does

  • An issue ledger — every error group the dashboard already fingerprints, plus the decision your team made about it: open, resolved, ignored, snoozed. A resolved group that fires again is a regression, and says so.
  • Incident correlation — when nine unrelated exceptions start within the same two minutes, that is one event, not nine. The correlator groups them and names the shared thing that failed. See incidents.
  • Alerting — rules that watch error rate, latency, throughput or any metric, plus event rules that fire on a new issue or a new incident, with cooldowns so one broken thing pages you once. See alerting.
  • Digests — a ranked list of what is worth looking at, and a Markdown brief you can hand to a coding assistant with the repository open. See digests.

Documentation

Next → Quickstart