Documentation
¶
Overview ¶
e2etiming publishes what a lifecycle run spent its time on.
The harness writes two files: a ledger of stages, one short JSON object per append, and the run's identity, written once. This joins them into one machine-readable report whose every record carries the identity a reader needs to compare two runs, and renders the same data as the job summary.
It also answers the two questions a duration alone cannot. Whether a run waited for a runner rather than working is read from the workflow's own job records; whether it was short of CPU, memory or disk is read from samples taken while it ran. Neither is inferred from the total.
Click to show internal directories.
Click to hide internal directories.