emrg: TUI daemon spawn throttle — complete the anti-storm fix by argszero · Pull Request #594 · argszero/emrg

argszero · 2026-08-09T09:05:45Z

Summary

Completes the Windows v0.2.15 storm fix (rant 2026-08-09T13:16:36) by closing the TUI client's respawn storm path — the last remaining piece of rant acceptance item ① ("启动 GUI/TUI 后零 cmd 窗口弹出").

PR #592 (CREATE_NO_WINDOW + GUI spawn throttle + scheduler backoff) and PR #593 (port-file self-heal + G43 PID guard) cover the daemon, scheduler, and GUI. But emrg/client/app.py _reconnect() had the same pattern the GUI had:

while True:
    await asyncio.sleep(1)
    conn = await daemon_manager.ensure_connected()   // daemon down → start_daemon() each second

With a down daemon, every iteration spawns a new daemon process every 1 second — on Windows each spawn was a cmd-window source (pre-#592); even windowless it is process churn + PID-lock exit races (post-#592).

Changes

`emrg/client/daemon_manager.py`

Module-level spawn throttle (mirrors GUI MAX_SPAWN_ATTEMPTS): max 3 spawn attempts per connect lifecycle, then start_daemon() raises a clear daemon failed to start after 3 attempts — please run 'emrg server' manually error instead of spawning forever.
Counter resets to 0 in ensure_connected() after a successful connect (same semantics as the GUI's auth-success reset).

`emrg/client/app.py`

_reconnect(): on throttle, surface a one-time system message + status hint (daemon down — run 'emrg server') so the host knows to intervene. Recovery path stays intact: host starts the daemon → is_running() true → connect succeeds → counter resets → loop returns.

Tests

Python 641 → 643 (tests/test_daemon_manager.py): start_daemon spawns exactly 3 times then throws the throttle error (4th call does not spawn — asserted via await_count); counter resets to 0 after a successful ensure_connected. Module-level autouse fixture prevents cross-test state leakage.
Doc counts synced (README.md / README.cn.md / Agent.md, emrg: sync test counts to 484 + guard test against doc drift (recurs #426/#430/#510) #511 guard). GUI tests untouched (96).

Verification

uv run pytest tests/: 643 passed
Import + --help OK
Recovery semantics: throttle only blocks spawns, not connect retries — a manually-started daemon reconnects normally.

…026-08-09T13:16:36) PRs #592 (CREATE_NO_WINDOW + GUI spawn throttle + scheduler backoff) and #593 (port-file self-heal + G43 PID guard) contain the Windows v0.2.15 storm fix, but the TUI client had the same storm pattern the GUI had: app.py _reconnect() loops every 1s calling ensure_connected() → with a down daemon each iteration calls start_daemon() → spawns a NEW daemon process every second. On Windows each spawn was a cmd-window source (pre-#592); even windowless it is process churn + PID-lock exit races post-#592. This completes rant acceptance item ① ("启动 GUI/TUI 后零 cmd 窗口弹出"): - daemon_manager.start_daemon: spawn throttle — max 3 attempts per connect lifecycle, then raise with a clear 'run emrg server manually' message instead of spawning forever; counter resets on successful connect in ensure_connected (mirrors GUI daemon_client.js MAX_SPAWN_ATTEMPTS). - app.py _reconnect: on throttle, surface a one-time system message + status hint so the host knows to start the daemon manually (recovery path intact: host starts daemon → is_running True → connect succeeds → counter resets). Tests: +2 (start_daemon throttles after 3 attempts, no 4th spawn; counter resets on success). 641→643 py, doc counts synced (#511).

pm25coder · 2026-08-09T09:08:53Z

Tested this PR on Windows (win32, Python 3.13.4):

argszero

✅ LGTM — cycle 2026-08-09T17:07 (Round 188)

Reviewed the full diff on the PR branch:

daemon_manager.start_daemon spawn throttle (_MAX_SPAWN_ATTEMPTS=3) with reset in ensure_connected on success mirrors the GUI's daemon_client.js semantics exactly — consistent anti-storm design across both clients.
app.py _reconnect() surfaces the throttle as a one-time system message + status hint (daemon down — run 'emrg server') so the host knows to intervene; recovery path intact (host starts daemon → connect succeeds → counter resets → loop returns).
Tests cover both states: 3 spawns then no 4th (await_count asserted), counter reset on successful connect; module-level autouse fixture prevents cross-test leakage.

Verified on the PR branch: uv run pytest tests/ → 643 passed; import + --help OK. GUI tests untouched (96). Doc counts synced to the branch base (#511).

… 2026-08-09T13:16:36) After the spawn throttle exhausts, scheduleReconnect keeps retrying with backoff capped at 60s — each retry hit the throttle and re-sent daemon_stopped status, so the renderer appended a duplicate 'run emrg server' system message every minute forever. Add daemonStoppedNotified: the warning is sent once per connect lifecycle (reset on successful connect), symmetric with the TUI _throttle_warned guard (PR #594).

argszero

✅ LGTM — cycle 2026-08-09T17:22 (Round 191)

Second independent review of the TUI spawn throttle (re-verified fresh this cycle):

daemon_manager _MAX_SPAWN_ATTEMPTS=3 + _spawn_attempts — mirrors the GUI daemon_client.js semantics exactly (module-level counter, reset on successful connect in ensure_connected). Consistent anti-storm design across both clients.
app.py _reconnect() surfaces the throttle as a one-time system message + status hint (daemon down — run 'emrg server'); recovery path intact (host starts daemon → connect succeeds → counter resets → loop returns).
Message-match verified: raise "daemon failed to start after N attempts" ↔ _reconnect catches "failed to start after" — exact alignment.
Tests cover both states: 3 spawns then no 4th (await_count asserted), counter reset on success; module-level autouse fixture prevents cross-test leakage.

Fresh verification on the PR branch: uv run pytest tests/ → 643 passed; import + --help OK. GUI tests untouched (96). Doc counts synced to the branch base (#511).

…egression hotfix) (#592) * emrg: fix Windows cmd-window storm + daemon spawn throttle (v0.2.15 regression hotfix) Rant 2026-08-09T13:16:36 (emergency, highest priority): Windows v0.2.15 host observed continuous cmd popups (had to reboot) + daemon startup failure leaving GUI/scheduler unable to connect. Root causes: 1. Zero CREATE_NO_WINDOW anywhere in the codebase — every subprocess spawn (git/gh/bash tool/scheduler/daemon spawn) popped a console window on Windows. 2. GUI reconnect loop respawned emrgd.cmd every ~5s forever; each spawn raced the previous daemon's startup (G43 stale-port unlink deleted a healthy daemon's port file → its scheduler logged 'cannot connect' 93 times while the GUI kept spawning). 3. Scheduler had no connect-failure backoff. Fixes: - NEW emrg/_win.py: win32_no_window_kwargs() = {creationflags: CREATE_NO_WINDOW} on Windows, {} elsewhere. Splatted into all 34 subprocess call sites (bash_tool, scheduler x12, daemon x5, git_utils, daemon_manager, installer, __main__ x3, client/app x9). - GUI daemon_client.js: spawn throttle (max 3 attempts per connect lifecycle, then throw with emrgd.log tail instead of respawning); reset counter on successful auth. Both spawn-timeout errors now surface the real emrgd.log tail (readLogTail). - GUI main.js: reconnect exponential backoff 1s→2s→4s→…cap 60s; daemon_stopped status surfaces the real failure to the renderer (zh/en i18n) instead of infinite 5s respawns. - scheduler.py: connect-failure exponential backoff max(30s, interval*2^n) capped at 10 min — no more per-tick retry storm while the daemon is down. Tests: +6 Python (win32_no_window_kwargs POSIX/Windows/splat; backoff zero/exp/cap/floor; 641→647), +2 GUI (spawn throttle + counter reset; 96→98). Docs synced (#511 guard). macOS/Linux unaffected (win32 branch is a no-op empty dict). * emrg: GUI daemon_stopped warning one-time per connect lifecycle (rant 2026-08-09T13:16:36) After the spawn throttle exhausts, scheduleReconnect keeps retrying with backoff capped at 60s — each retry hit the throttle and re-sent daemon_stopped status, so the renderer appended a duplicate 'run emrg server' system message every minute forever. Add daemonStoppedNotified: the warning is sent once per connect lifecycle (reset on successful connect), symmetric with the TUI _throttle_warned guard (PR #594).

pm25coder · 2026-08-09T09:30:47Z

Post-merge conflict check on this Windows host after #592 merged to master (now 9699135):

argszero

✅ LGTM — cycle 2026-08-09T17:30 (Round 192)

Reviewed the branch diff against current master (post-#593 merge):

_MAX_SPAWN_ATTEMPTS=3 guard in daemon_manager.start_daemon is the client-side total gate for the v0.2.15 cmd-storm root cause — it caps respawns per connection lifecycle and surfaces a clear RuntimeError when exhausted; counter resets on ensure_connected success (verified the reset only happens after a successful connect).
app.py _reconnect catches the throttle error once (_throttle_warned) and shows a one-time system message + 'daemon down — run emrg server' status hint — correct user-facing degradation instead of silent 1s respawn loop.
Positive/negative states covered by tests: 3 spawns then no 4th (await_count asserted), counter reset on successful connect, autouse fixture prevents cross-test leakage. tests/test_daemon_manager.py 29 passed locally.
Matches GUI daemon_client.js spawn-throttle semantics (#593 branch) — consistent anti-storm design across both clients.

Resolve doc-count conflicts after #592+#593 merged: 650+2=652 py, 101 gui (take highest). Production code merged cleanly.

pm25coder · 2026-08-09T09:34:27Z

Re-verified the rebased branch (765822e on top of f91298f) on this Windows host:

…e self-heal + #594 TUI spawn throttle entries (#595) Co-authored-by: EMRG Evolution <emrg@argszero.dev>

…T13:16:36) (#596) Carries the emergency Windows v0.2.15 regression hotfix (host observed cmd-window storm + daemon startup failure, had to reboot): - #592: CREATE_NO_WINDOW across all 34 subprocess sites (emrg/_win.py win32_no_window_kwargs) + GUI spawn throttle + scheduler connect backoff - #593: daemon port-file self-heal (_port_keepalive_loop) + G43 PID guard (never delete a live daemon's port file) - #594: TUI spawn throttle (_MAX_SPAWN_ATTEMPTS=3 + reset on connect) - #595: evolution_prompt quick-ref entries Per host authorization (rant #16/#18: release without waiting for real-machine verification), this bump goes straight to Build Release on merge.

argszero commented Aug 9, 2026

View reviewed changes

argszero mentioned this pull request Aug 9, 2026

emrg: fix Windows cmd-window storm + daemon spawn throttle (v0.2.15 regression hotfix) #592

Merged

argszero commented Aug 9, 2026

View reviewed changes

EMRG Evolution added 2 commits August 9, 2026 17:31

emrg: resolve doc test-count conflicts with master (650/101)

6a99a51

emrg: sync doc test counts to 652 (post-#593 merge + #594 tests)

765822e

argszero merged commit e6468fd into master Aug 9, 2026

argszero added a commit that referenced this pull request Aug 9, 2026

merge origin/master into feature/tui-spawn-throttle (#594)

3e0b495

Resolve doc-count conflicts after #592+#593 merged: 650+2=652 py, 101 gui (take highest). Production code merged cleanly.

argszero added a commit that referenced this pull request Aug 9, 2026

fix doc test counts to 652 py after #592+#593 merge (#594 rebase)

250960b

argszero mentioned this pull request Aug 9, 2026

emrg: evolution_prompt quick-ref — add #592/#593/#594 anti-storm fix entries #595

Merged

argszero added a commit that referenced this pull request Aug 9, 2026

emrg: evolution_prompt quick-ref — add #592 cmd-storm + #593 port-fil…

dfc792c

…e self-heal + #594 TUI spawn throttle entries (#595) Co-authored-by: EMRG Evolution <emrg@argszero.dev>

argszero mentioned this pull request Aug 9, 2026

emrg: bump v0.2.16 — Windows cmd-window storm hotfix (rant 2026-08-09T13:16:36) #596

Merged

argszero mentioned this pull request Aug 9, 2026

emrg: evolution_prompt quick-ref — add #600 GUI streaming markdown block projection entry #601

Merged

Navigation Menu

emrg: TUI daemon spawn throttle — complete the anti-storm fix - #594

emrg: TUI daemon spawn throttle — complete the anti-storm fix#594
argszero merged 3 commits into
masterfrom
feature/tui-spawn-throttle

argszero commented Aug 9, 2026

Uh oh!

pm25coder commented Aug 9, 2026

Uh oh!

argszero left a comment

Uh oh!

argszero left a comment

Uh oh!

pm25coder commented Aug 9, 2026

Uh oh!

argszero left a comment

Uh oh!

pm25coder commented Aug 9, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Conversation

argszero commented Aug 9, 2026

Summary

Changes

emrg/client/daemon_manager.py

emrg/client/app.py

Tests

Verification

Uh oh!

pm25coder commented Aug 9, 2026

Uh oh!

argszero left a comment

Choose a reason for hiding this comment

Uh oh!

argszero left a comment

Choose a reason for hiding this comment

Uh oh!

pm25coder commented Aug 9, 2026

Uh oh!

argszero left a comment

Choose a reason for hiding this comment

Uh oh!

pm25coder commented Aug 9, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

`emrg/client/daemon_manager.py`

`emrg/client/app.py`