Expand description
Logging dataflow for the resource usage of the replica’s processes.
The observations are read by mz_metrics::usage, on a task that keeps sampling while the
timely workers are busy. This dataflow only reports them, so a worker that stalls delays the
report without losing an observation: the values it reads are either kernel-maintained
high-water marks or the sampler’s own folded peaks, neither of which a late read can miss.
One row per (process_id, source, metric), so a metric that moves every sample does not drag
the stable ones through a retraction with it.