Summary

• PR fix(stall): bound the verify retrieve separately from the create (X1D re-ship loop) #37 fixed ship_object's (size-routing) retrieve sharing the create-side HTAR_STALL_SECONDS (2h) silence budget. send_to_fortress — the whole-target and Phase-3 leveled-incremental transport — carries its own, fully self-contained inline twin of the stall-watch logic (run_watched, invariant chore(reclaim): hold AAPF landing-zone raw until promote loop verified #1: no outer-scope references so it stays safe under --globus) and was left on its own hardcoded 7200s default for both the create step and the two retrieve steps (hsi get, then local tar xvf).
• Confirmed independently hit on repository_X0H_2_spectral-standoff (job 41417829, 2026-07-31, a leveled-incremental, multi-TB whole-target tar): create succeeded, then the round-trip retrieve was killed at exactly "no output for 2.0h" — the same bug class PR fix(stall): bound the verify retrieve separately from the create (X1D re-ship loop) #37 fixed for the routed path.
• send_to_fortress now takes retrieve_stall_seconds and requires it (raises before any htar_large/hsi/tar call) — mirroring ship_object's contract exactly, no code-side fallback that could silently reinstate the 2h bound.
• Threaded only into the two retrieve run_watched() calls (hsi get, tar xvf); the create call is untouched (still the hardcoded 7200s default — a create streams per-file progress, so silence there really is a hang).
• Both real call sites (whole-target Step 2, and Phase-3's ship_level) updated to pass the already-loaded config value through positionally (run only forwards positional args).
• Design choice: reused stall_defaults.json's existing retrieve_stall_seconds default/config key rather than adding a second knob — it was already documented path-agnostically ("the post-create round-trip verify retrieve"). An operator can still raise it per-asset for X0H-sized multi-TB tars via the same override mechanism that already exists. Happy to split this into a separate default if review disagrees.
• CLAUDE.md, stall_defaults.json, and config.example.json comments updated to reflect both transports sharing this one bound.

Test plan

• New tests/test_send_to_fortress_retrieve_stall.py: fakes htar_large/hsi/tar as shell-script stand-ins on PATH (matching test_ship_object_htar_rc.py's approach), since run_watched is a closure local to send_to_fortress and can't be intercepted the way archive._run_with_stall_watch is for ship_object. Each behavioral assertion runs send_to_fortress on a background thread with a bounded join() so a regression that falls back to the 2h default fails fast (not a real 2-hour hang).
  • hsi get retrieve honors the configured bound, not the 2h default
  • tar xvf retrieve honors the configured bound, not the 2h default
  • create keeps its own hardcoded default even given a tiny configured retrieve bound
  • missing/invalid retrieve_stall_seconds raises before any tape work
• Full existing suite: 428 passed (423 before this change + 5 new), no regressions.

Not merging or deploying — flagging for review per usual process.

🤖 Generated with Claude Code