Logs, metrics, traces, incidents and understanding production systems.
Showing 1–3 of 3 dossiers
Sentry has completed a breaking alerting migration. Legacy alert APIs are gone; metric detection now lives in Monitors while notification routing lives in Alerts, and old direct integrations must use the replacement endpoints.
Grafana’s GA agent-observability stack can track latency, tokens, cost and conversations, score live traffic with deterministic or LLM-based evaluators, route failures into test collections, compare experiments and use those results as pull-request gates. Evaluator quality and instrumentation coverage remain the main limits.
The new tracing path gives Supabase requests a shared trace_id across browser or app spans and backend logs, with support for OpenTelemetry-compatible tooling. For JavaScript users, the practical detail is a required upgrade to supabase-js 2.112.0 or later because 2.106.0 through 2.111.x silently failed to propagate headers in bundled applications.
Observability helps a team explain what its software is doing before an incident becomes a customer support mystery. Logs, metrics and traces are useful only when they can be collected, queried and retained at a cost that matches the service. Tooling changes often affect both operational clarity and a surprisingly large bill.
BTN follows important developments in telemetry standards, monitoring platforms, incident tooling and observability pricing. Coverage asks whether a change improves diagnosis, reduces lock-in or simply creates more data to manage. It is written for builders who want enough signal to run reliable products without copying an enterprise stack. The useful outcome is knowing what to instrument, what to retain and where a vendor decision changes the economics.
OpenTelemetry and other shared formats matter when they make telemetry easier to move and correlate. BTN also watches sampling, retention and high-cardinality pricing, where apparently small configuration choices can dominate cost. Good observability should shorten an investigation rather than create a second system that needs its own investigation.