Create a new run for failed work

A retry does not resume scheduler jobs in place and does not change the original run. It creates a new run_project() submission containing the failed or cancelled stages and subjects found in the source run. The selected work is rerun even if old completion markers would normally skip it, and the new provenance record identifies the source run.

Usage

retry_project_run(
  input,
  run_id = "latest",
  include_blocked = FALSE,
  dry_run = TRUE
)

Arguments

input

A project configuration object, YAML file, or project directory.

run_id

Source run ID returned by run_project() or listed in inspect_project(input)$runs. An explicit ID is recommended for retry.

include_blocked

Also include downstream jobs marked FAILED_BY_EXT. These jobs did not fail themselves; they could not run because an earlier required job failed. They are excluded by default.

dry_run

If TRUE, return a plan without submitting jobs. If FALSE, submit the new run immediately.

Value

A bg_project_plan for a dry run or bg_project_run after submission.

Details

Preview with dry_run = TRUE before submitting. The preview returns a plan and contacts no scheduler. With dry_run = FALSE, submission begins immediately and the function returns the new run handle.

See also

diagnose_project() to inspect the source failure and get_run_provenance() to compare the original and retry runs.

Examples

if (FALSE) { # \dontrun{
# Inspect and correct the failure before retrying.
diagnosis <- diagnose_project(
  scfg, run$run_id, interactive = FALSE
)

# Preview only; no jobs are submitted.
retry_plan <- retry_project_run(scfg, run$run_id, dry_run = TRUE)

# Submit a separate new run after reviewing the plan.
retry_run <- retry_project_run(scfg, run$run_id, dry_run = FALSE)
} # }