cmd/archie-state-store TestStateStoreRealProcessSmoke/Deadline/RestartRecovery/RemoteSurfaces fail with 'did not recover bound addresses from log' and an empty log after the 15s poll. Reproduced on untouched HEAD (worktree at 5cba869), so it is not caused by the gateway-auth change. The binary itself works: run in the foreground it logs both bound addresses immediately and serves. Under the test's piped subprocess (and a detached redirect) it creates its SQLite files but emits zero stderr while staying alive. Suspect sandbox subprocess-pipe behavior, not the product code. Needs a host/CI run to tell a real test bug from a sandbox artifact.
Close reason
Sandbox artifact, not a test bug. Host run 2026-09-09 on the gateway-auth tree: 'go test ./cmd/archie-state-store/... -count=1' => ok, 17.5s, and a full 'task check' exits 0 including these tests. The same gate also passed clean on 5cba869 before the gateway-auth change existed. The empty-child-log symptom is specific to the restricted sandbox's subprocess pipe handling; reopen if it appears on a host or in CI.
Imported from beads archie-core-eqb1 · GitHub #782 · type task · priority P2 · created 2026-09-08 · closed 2026-09-08
cmd/archie-state-store TestStateStoreRealProcessSmoke/Deadline/RestartRecovery/RemoteSurfaces fail with 'did not recover bound addresses from log' and an empty log after the 15s poll. Reproduced on untouched HEAD (worktree at 5cba869), so it is not caused by the gateway-auth change. The binary itself works: run in the foreground it logs both bound addresses immediately and serves. Under the test's piped subprocess (and a detached redirect) it creates its SQLite files but emits zero stderr while staying alive. Suspect sandbox subprocess-pipe behavior, not the product code. Needs a host/CI run to tell a real test bug from a sandbox artifact.
### Close reason
Sandbox artifact, not a test bug. Host run 2026-09-09 on the gateway-auth tree: 'go test ./cmd/archie-state-store/... -count=1' => ok, 17.5s, and a full 'task check' exits 0 including these tests. The same gate also passed clean on 5cba869 before the gateway-auth change existed. The empty-child-log symptom is specific to the restricted sandbox's subprocess pipe handling; reopen if it appears on a host or in CI.
---
_Imported from beads `archie-core-eqb1` · GitHub #782 · type task · priority P2 · created 2026-09-08 · closed 2026-09-08_
sam
added the mediumtask labels 2026-10-27 13:39:52 +00:00
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
cmd/archie-state-store TestStateStoreRealProcessSmoke/Deadline/RestartRecovery/RemoteSurfaces fail with 'did not recover bound addresses from log' and an empty log after the 15s poll. Reproduced on untouched HEAD (worktree at 5cba869), so it is not caused by the gateway-auth change. The binary itself works: run in the foreground it logs both bound addresses immediately and serves. Under the test's piped subprocess (and a detached redirect) it creates its SQLite files but emits zero stderr while staying alive. Suspect sandbox subprocess-pipe behavior, not the product code. Needs a host/CI run to tell a real test bug from a sandbox artifact.
Close reason
Sandbox artifact, not a test bug. Host run 2026-09-09 on the gateway-auth tree: 'go test ./cmd/archie-state-store/... -count=1' => ok, 17.5s, and a full 'task check' exits 0 including these tests. The same gate also passed clean on 5cba869 before the gateway-auth change existed. The empty-child-log symptom is specific to the restricted sandbox's subprocess pipe handling; reopen if it appears on a host or in CI.
Imported from beads
archie-core-eqb1· GitHub #782 · type task · priority P2 · created 2026-09-08 · closed 2026-09-08