Skip to contents

Plot the last-run duration of every stored target from an autometric log

Source: R/ga_autometric_runs.R
ga_autometric_lastrun.Rd

lifecycle-experimental

Reconstruct the resource impact of the last pipeline run from an autometric log. Each (pid, phase) execution is collapsed to its duration and mean CPU usage, the log is restricted to phases that are still present in the targets store (targets::tar_objects()), and for each target only the single longest run is kept. The result is one bar per stored target, mapping duration to bar length and mean CPU usage to fill.

This is the log-based companion of ga_targets_meta_plot(): the former reads the metadata, the latter re-derives the same "last run" picture from the finer-grained resource samples. To visualize every run instead of only the last, use ga_autometric_history().

Usage

ga_autometric_lastrun(
  log,
  store = targets::tar_config_get("store"),
  object_names = NULL,
  units_time = "hours",
  units_memory = "gigabytes",
  ...
)

Arguments

log

Either a path to an autometric log file (read with autometric::log_read()) or a data frame already returned by autometric::log_read() (with at least the columns pid, phase, time, cpu).

store

(character, default targets::tar_config_get("store")) Path to the targets data store, used to list the objects still present via targets::tar_objects().

object_names

(optional character) Vector of target names to keep. When supplied, store is not queried. Useful for tests or to focus on a subset.

units_time, units_memory

Passed to autometric::log_read() when log is a file path (defaults "hours" and "gigabytes").

...

Additional arguments forwarded to autometric::log_read().

Value

A ggplot object with one bar per stored target.

See also

ga_targets_meta_plot(), ga_autometric_history(), ga_autometric_plot()

Author

Adrien Taudière

Examples

if (requireNamespace("autometric", quietly = TRUE)) {
  log_df <- data.frame(
    version = "0.1.2",
    phase = rep(c("raw", "scaled"), each = 4),
    pid = rep(c(1L, 2L), each = 4),
    name = "local",
    status = 0L,
    time = c(0, 0.1, 0.2, 0.3, 0.3, 0.35, 0.4, 0.45),
    core = runif(8, 0, 40),
    cpu = runif(8, 0, 90),
    resident = runif(8, 70, 90),
    virtual = runif(8, 900, 1000)
  )
  ga_autometric_lastrun(log_df, object_names = c("raw", "scaled"))
}