Stage State Retention and Advanced Re-Run
pending feedback
C
Civic Parrotfish
Allow Harness to retain the state of previous stages, enabling users to make decisions such as re-running only failed test cases. Users should have the ability to re-run failed stages or tests directly from the UI, or via an out-of-the-box solution
Log In
S
Shamrock green Newt
Adding some more context on this
Enable Harness to retain execution state from previous pipeline stages, allowing users to re-run only failed test cases rather than re-executing entire stages. This includes:
Selective Test Re-Run: Identify and re-run only failed tests from a previous execution while preserving successful test results
UI Controls: One-click re-run of failed tests directly from the pipeline execution Tests tab
API/OOTB Support: Native API endpoints and step templates without requiring custom scripting
Aggregated Results: Unified view combining original and re-run results for accurate pipeline status
Business Impact
Cost Reduction: Eliminates redundant test executions when addressing flaky tests, reducing CI/CD compute costs by 20-40% for organizations with large test suites.
Developer Productivity: Reduces feedback loop from 60+ minutes (full suite re-run) to under 5 minutes for failed test verification, recovering 2-4 hours per developer weekly.
Competitive Parity: CircleCI, GitLab, and other platforms offer "re-run failed tests only" capabilities. This gap is a common objection during Harness CI evaluations and migration discussions.
Pipeline Velocity: Faster mean-time-to-green on PRs, reduced queue congestion, and accelerated release cycles.
s
shivkumar.loka
marked this post as
pending feedback
s
shivkumar.loka
Hey Kartik,
Currently we have a re-run pipeline from failed stage or a specific stage option (see screenshot referenced below). Tests (legacy run tests or test intelligence step) is scoped to a step and unfortunately that is not an option available at the step level. We're working on revamping this experience and will have a solution for this soon. Please stay tuned for updates.
Thank you
Regards
Shivkumar Loka
Harness Product Team
C
Civic Parrotfish
As I understand it, if a previously failed stage depends on another stage, that dependent stage should also be re-run automatically.
However, in some cases — for example, if I need to re-run only the Test stage — I may not want to re-run the Build stage again, since I already know the issue lies in the Test stage. So i would like to stage should retain the stag data to re run only the particular stage .