Observability

Observability is the set of signals that each layer of the stack produces, from host metrics to agent traces, and diagnosing a failure usually means joining them.

Observability is the set of signals that each layer of the stack produces. The signals follow the shape of the stack. Host and hardware metrics are at the bottom, per-sandbox resource usage and logs are in the middle, and agent traces are at the top. Diagnosing a failure usually means joining the signals of several layers.