You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Control harness setup exposes its temporary write token in jq arguments #554
The existing ordinary phase_control in tests/e2e-pithead-control.sh interpolates its temporary ACCESS_TOKEN into the jq program passed to set_cfg. The shared set_cfg implementation in tests/e2e-pithead.sh invokes jq "$1", so that live write token becomes a process argument visible to a local process observer during setup. Passing HTTP bearer headers through stdin does not protect this earlier setup step.
Expected: credential-bearing config setup uses private stdin or an owner-only file/descriptor with a constant jq program; process arguments and logs carry no token. Add a hardware-free regression inspecting jq arguments. Preserve setup/restoration behavior and ordinary gate coverage.
This pattern predates #540. The new queued replay setup avoids it, and this side issue owns correction of ordinary control setup rather than enlarging the producer-latency investigation. No real token or private configuration was read or published.
Blocked on bench-ci#1063. Job 2016@1ff6628b056227907a3158fd62ec0cf217133e3d ended in error: preflight refused the unverified live-stack baseline, so the candidate control scenarios never ran. Full status/log and the empty artifact inventory were read; no rows ran, rig restoration succeeded, and baseline verification failed. This supplies no hardware verdict for #554.
The code, regression tests, corrected documentation and author-side adversarial/security reviews are complete in draft PR #560, and all eight GitHub checks pass at that head. The managed recovery role owning bench-ci#1063 must diagnose and restore current baseline health and verify safe admission. Once that prerequisite closes, this issue needs a fresh narrow ordinary control job on the current head and review of its full log, artifacts and exact evidence. No blind retry or manual bench/rig action was performed.
Blocked on p2pool-starter-stack/bench-ci#1063. Blocked on bench-ci#1063: job 2016@1ff6628b056227907a3158fd62ec0cf217133e3d ended in error before control scenarios ran because the baseline was unverified. The fix, regressions, documentation, author-side reviews and all eight GitHub checks are complete; managed recovery must restore and verify the baseline before a fresh control proof. Exact evidence and blocker comments are posted on the PR and both issues.
The runner resumes this issue when that one closes.
Activity
pithead-developer commentedon Oct 2, 2026
Blocked on bench-ci#1063. Job
2016@1ff6628b056227907a3158fd62ec0cf217133e3dended in error: preflight refused the unverified live-stack baseline, so the candidate control scenarios never ran. Full status/log and the empty artifact inventory were read; no rows ran, rig restoration succeeded, and baseline verification failed. This supplies no hardware verdict for #554.The code, regression tests, corrected documentation and author-side adversarial/security reviews are complete in draft PR #560, and all eight GitHub checks pass at that head. The managed recovery role owning bench-ci#1063 must diagnose and restore current baseline health and verify safe admission. Once that prerequisite closes, this issue needs a fresh narrow ordinary
controljob on the current head and review of its full log, artifacts and exact evidence. No blind retry or manual bench/rig action was performed.pithead-developer commentedon Oct 2, 2026
Blocked on p2pool-starter-stack/bench-ci#1063. Blocked on bench-ci#1063: job 2016@1ff6628b056227907a3158fd62ec0cf217133e3d ended in error before control scenarios ran because the baseline was unverified. The fix, regressions, documentation, author-side reviews and all eight GitHub checks are complete; managed recovery must restore and verify the baseline before a fresh control proof. Exact evidence and blocker comments are posted on the PR and both issues.
The runner resumes this issue when that one closes.
pithead-developer commentedon Oct 2, 2026
Merged in #560.