`orca.estimate`

How long runs of a pipeline would take, from how long its recent runs took.

A pipeline's last SAMPLE succeeded runs give its seconds per partition: their running time (started to ended, not the wait in the queue) over the partitions they ran. A backfill of runs runs over keys partitions takes waves of max_runs runs at once, each about as long as a run of its size; one run over them all takes them one after another.

Structs

`Pace`

struct Pace

A pipeline's seconds per partition, and the runs that says it.

Implements Copyable, Movable

Fields: per_key: Float64: -1 when no run has succeeded yet.; runs: Int

`known`

def known(self) -> Bool

`backfill`

def backfill(self, keys: Int, runs: Int, max_runs: Int) -> Int

Seconds for runs runs over keys partitions, max_runs at once; -1 if not known.

`one_run`

def one_run(self, keys: Int) -> Int

Seconds for one run over keys partitions; -1 if not known.

Functions

`pace`

def pace(mut db: State, pipeline: String) -> Pace

pipeline's pace over its last SAMPLE succeeded runs.

`per_key_text`

def per_key_text(p: Pace, unit: String) -> String

"95 s a day over its last 20 runs".

`estimate_text`

def estimate_text(p: Pace, seconds: Int, unit: String, at_once: Int) -> String

"About 2h 10m, 1 run at once (95 s a day over its last 20 runs)." or why there's no estimate.

Aliases

comptime SAMPLE = 20

The most recent succeeded runs the estimate is taken from.