Riemann

A stream of individual monitoring events may need interpretation before it becomes a useful alert. Riemann accepts events from systems and applications, then processes them through programmable streams that can filter, combine and route observations. Teams can express behavior such as tracking temporary state or notifying on a particular event sequence. Its rules are executable Clojure code, so adopting or migrating it requires understanding that logic rather than treating the configuration as only a list of monitoring targets.

Current guidance

Its own guide describes Clojure stream functions, an event index and configuration reload behavior. That differs from simply scraping numeric time series: existing deployments may encode filtering, correlation, expiration and notification logic inside their stream configuration.

Choose Riemann when that event-processing model fits the team’s needs and skills, and review the selected JVM, application package and integrations independently of the archived chart. Treat configuration as executable code with access to the process environment. Limit inbound event sources and keep notification credentials out of unreviewed templates or broadly readable logs.

Before changing the image or moving logic elsewhere, save configuration, dependency versions, input schemas and expected outputs for representative events. Replay a synthetic sequence including expiry, duplicates and out-of-order input, then verify notification behavior without contacting real recipients. Upstream notes reload caveats and retention of old configuration after a syntax error, so confirm the active configuration explicitly. No current package-maintenance guarantee or production event-pipeline migration is claimed here.