Non-Fail-Fast and Recovery¶
Choose between strict fail-fast execution and best-effort continuation.
When to use¶
Use this when deciding how runtime should behave after stage/plugin failures.
Prerequisites¶
- A pipeline config with
[runtime]
Steps¶
Set strict mode:
[runtime]
fail_fast = true
random_seed = 0
Set recovery mode:
[runtime]
fail_fast = false
random_seed = 0
Run and inspect diagnostics in recovery mode:
$ honestroles run --pipeline-config pipeline.toml --plugins plugins.toml
Expected result¶
fail_fast = true: execution stops at firstHonestRolesError.fail_fast = false: execution continues, and diagnostics include:
{
"non_fatal_errors": [
{
"stage": "filter",
"error_type": "PluginExecutionError",
"detail": "..."
}
]
}
Warning
Recovery mode can produce partial outputs. Always inspect non_fatal_errors before using results.
Next steps¶
- Failure taxonomy and wrappers: Error Model
- Practical failure signatures: Common Errors