fix(runs): every PR run re-establishes its checkout inside the retryable step - #130
Merged
Merged
Conversation
…ble step #128 gave `offload-test`'s staged path a checkout probe and deliberately left the single-exec runs alone — their contract was pinned and their exposure was one step. The evidence disagreed within the hour: `check` posted ExecFailed: exec failed (exit -1): working directory '/workspace/<repo>' was missing at exec time — the checkout did not survive to this step (container recycled) as a red verdict on a 40-second run, with all three platform retries spent re-running the command in the same absent directory. So the probe becomes a primitive — `ensureWorkspace` beside `workspace` — and `check`, `oxlint` and `offload-test` all call it inside the step that retries. One `test -d` on the happy path; a clone and an install on a recycled container, which is what the step needed anyway. `offload-test`'s local copy is deleted in favour of the shared one. Exposure is not one step: a container can be recycled between ANY two durable steps, and the checkout is a step earlier than every exec by construction.
![flaredispatch-fractalboxdev[bot]](https://avatars.githubusercontent.com/in/4273665?s=60&v=4){kind=small}
This was referenced Aug 15, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Problem & Insight
#128 gave
offload-test's staged path a checkout probe and deliberately left the single-exec runs alone — "its contract is pinned byte for byte, and its exposure is one step rather than five".That reasoning was wrong, and the evidence arrived within the hour. A consumer's
checkposted this as a red verdict on a 40-second run:All three platform retries were spent re-running the command in the same absent directory. Exposure is not "one step": a container can be recycled between any two durable steps, and
checkoutis a step earlier than every exec by construction. A single-exec run has exactly the same gap as a staged one — it just has fewer chances to notice.Take
The probe becomes a primitive —
ensureWorkspacebesideworkspace— andcheck,oxlintandoffload-testall call it inside the step that retries. That placement is the whole point: the rebuild has to be part of what a retry re-runs, or the retry's precondition is the thing the failure destroyed.test -d, about a second.offload-test's local copy from fix(offload-test): a stage probes for its checkout before it runs #128 is deleted in favour of the shared one.Key actions
check.test.tspins the recovery: probe answers non-zero → two clones, command still runs, run goes greenoffload-test.test.tskeeps its equivalent case, now against the primitive