// the find
postmates/cernan
telemetry aggregation and shipping, last up the ladder
Cernan is a telemetry aggregation daemon written in Rust that ingests metrics and logs from multiple protocols (statsd, graphite, collectd, Avro) and ships them to multiple backends simultaneously. It supports Lua scripting for in-flight transformation and was built by Postmates for internal use. The repo is archived and unmaintained — the fork at github.com/blt/cernan is where any activity happens now.
Protocol breadth is real: it handles statsd, graphite, Prometheus, native protobuf, and Avro as sources, and can fan out to InfluxDB, Elasticsearch, Kafka, Wavefront, and Prometheus simultaneously. The Lua filter layer is genuinely useful — you can reshape, drop, or derive new metrics without recompiling. Written in Rust so memory and CPU footprint are low, which matters when you're running this on every host. The TOML config model with named sources, filters, and sinks wired together is cleaner than most OSS telemetry tools of the same era.
The repo is archived. Postmates itself no longer exists. The fork is one person's personal repo with no apparent community. Adopting this means you own the Rust codebase from day one. OpenTelemetry has eaten this entire problem space since 2019 — there's no OTLP source or sink, so cernan can't participate in any modern observability stack. The Lua scripting dependency adds a runtime escape hatch that's hard to test and debug in production. 314 stars and 10 forks over ~7 years of life is a signal about how much real-world adoption it found outside of Postmates.