From 403af92504174bae6adda292458a3f730914e7ea Mon Sep 17 00:00:00 2001
From: LeonSGP43 <cine.dreamer.one@gmail.com>
Date: Wed, 27 May 2026 11:18:01 +0800
Subject: [PATCH] fix(gateway): avoid restart loops on replace

---
 hermes_cli/gateway.py                    | 14 ++++++++------
 tests/hermes_cli/test_gateway_service.py |  4 ++++
 2 files changed, 12 insertions(+), 6 deletions(-)

diff --git a/hermes_cli/gateway.py b/hermes_cli/gateway.py
index 867319574806..6d59fe127e29 100644
--- a/hermes_cli/gateway.py
+++ b/hermes_cli/gateway.py
@@ -207,9 +207,11 @@ def _graceful_restart_via_sigusr1(pid: int, drain_timeout: float) -> bool:
 
     SIGUSR1 is wired in gateway/run.py to ``request_restart(via_service=True)``
     which drains in-flight agent runs (up to ``agent.restart_drain_timeout``
-    seconds), then exits with code 75.  Both systemd (``Restart=always``
-    + ``RestartForceExitStatus=75``) and launchd (``KeepAlive.SuccessfulExit
-    = false``) relaunch the process after the graceful exit.
+    seconds), then exits with code 75. Systemd units generated by Hermes use
+    ``Restart=on-failure`` together with ``RestartForceExitStatus=75`` so the
+    service is relaunched after the graceful exit without reviving clean
+    ``--replace`` takeovers. launchd still uses ``KeepAlive.SuccessfulExit =
+    false`` for the same relaunch behavior.
 
     This is the drain-aware alternative to ``systemctl restart`` / ``SIGTERM``,
     which SIGKILL in-flight agents after a short timeout.
@@ -2217,7 +2219,7 @@ def generate_systemd_unit(system: bool = False, run_as_user: str | None = None)
 Environment="PATH={sane_path}"
 Environment="VIRTUAL_ENV={venv_dir}"
 Environment="HERMES_HOME={hermes_home}"
-Restart=always
+Restart=on-failure
 RestartSec=5
 RestartMaxDelaySec=300
 RestartSteps=5
@@ -2252,7 +2254,7 @@ def generate_systemd_unit(system: bool = False, run_as_user: str | None = None)
 Environment="PATH={sane_path}"
 Environment="VIRTUAL_ENV={venv_dir}"
 Environment="HERMES_HOME={hermes_home}"
-Restart=always
+Restart=on-failure
 RestartSec=5
 RestartMaxDelaySec=300
 RestartSteps=5
@@ -3236,7 +3238,7 @@ def run_gateway(verbose: int = 0, quiet: bool = False, replace: bool = False):
     print()
     
     # Exit with code 1 if gateway fails to connect any platform,
-    # so systemd Restart=always will retry on transient errors
+    # so systemd Restart=on-failure will retry on transient errors
     verbosity = None if quiet else verbose
 
     # ── Exit-path diagnostics ────────────────────────────────────────────
diff --git a/tests/hermes_cli/test_gateway_service.py b/tests/hermes_cli/test_gateway_service.py
index b1fcadbf4f0d..55540fb8f5dd 100644
--- a/tests/hermes_cli/test_gateway_service.py
+++ b/tests/hermes_cli/test_gateway_service.py
@@ -326,6 +326,8 @@ def test_user_unit_avoids_recursive_execstop_and_uses_extended_stop_timeout(self
         assert "ExecStart=" in unit
         assert "ExecStop=" not in unit
         assert "ExecReload=/bin/kill -USR1 $MAINPID" in unit
+        assert "Restart=on-failure" in unit
+        assert "Restart=always" not in unit
         assert f"RestartForceExitStatus={GATEWAY_SERVICE_RESTART_EXIT_CODE}" in unit
         # TimeoutStopSec must exceed the default drain_timeout (60s) so
         # systemd doesn't SIGKILL the cgroup before post-interrupt cleanup
@@ -387,6 +389,8 @@ def test_system_unit_avoids_recursive_execstop_and_uses_extended_stop_timeout(se
         assert "ExecStart=" in unit
         assert "ExecStop=" not in unit
         assert "ExecReload=/bin/kill -USR1 $MAINPID" in unit
+        assert "Restart=on-failure" in unit
+        assert "Restart=always" not in unit
         assert f"RestartForceExitStatus={GATEWAY_SERVICE_RESTART_EXIT_CODE}" in unit
         # TimeoutStopSec must exceed the default drain_timeout (60s) so
         # systemd doesn't SIGKILL the cgroup before post-interrupt cleanup