fix: bounded process log rotation and timestamp-sortable IDs (#52, #54)

Co-Authored-By: wjss <a123edcrfv@live.cn>

Author: tjbck

CHANGELOG.md

@@ -4,6 +4,13 @@ All notable changes to this project will be documented in this file.
 
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/).
 
+## [0.11.10] - 2026-03-11
+
+### Fixed
+
+- 🧟 **Zombie process cleanup** — process runner `kill()` methods now use `os.killpg()` to signal the entire process group instead of just the direct child PID. Background processes started inside terminals or `/execute` sessions (e.g. `sleep 100 &`) are now properly terminated on cleanup. `_cleanup_session()` always calls `process.wait()` after force-killing to prevent zombie entries in the process table.
+- 🐳 **Docker PID 1 reaping** — added `tini` as the container's init process (`ENTRYPOINT ["/usr/bin/tini", "--", ...]`). Python/uvicorn no longer runs as PID 1, so orphaned grandchild processes are automatically reaped instead of accumulating as zombies.
+
 ## [0.11.9] - 2026-03-11
 
 ### Added

Dockerfile

@@ -23,7 +23,7 @@ RUN apt-get update && apt-get install -y --no-install-recommends \
     zip unzip tar gzip bzip2 xz-utils zstd p7zip-full \
     # System
     procps htop lsof strace sysstat \
-    sudo tmux screen \
+    sudo tmux screen tini \
     ca-certificates gnupg apt-transport-https \
     # Capabilities (needed for setcap on Python binary)
     libcap2-bin \
@@ -67,5 +67,5 @@ EXPOSE 8000
 
 COPY entrypoint.sh /app/entrypoint.sh
 
-ENTRYPOINT ["/app/entrypoint.sh"]
+ENTRYPOINT ["/usr/bin/tini", "--", "/app/entrypoint.sh"]
 CMD ["run"]

open_terminal/main.py

@@ -1246,7 +1246,13 @@ async def port_proxy(port: int, path: str, request: Request):
 
 
     def _cleanup_session(session_id: str):
-        """Clean up a terminal session's resources."""
+        """Clean up a terminal session's resources.
+
+        For PTY sessions the shell is spawned with ``start_new_session=True``,
+        giving it a dedicated process group.  We signal the *entire* group so
+        that background jobs started inside the terminal (e.g. ``sleep 999 &``)
+        are also reaped, and always call ``process.wait()`` to avoid zombies.
+        """
         session = _terminal_sessions.pop(session_id, None)
         if session is None:
             return
@@ -1258,13 +1264,23 @@ def _cleanup_session(session_id: str):
                 os.close(session["master_fd"])
             except OSError:
                 pass
+
             process = session["process"]
             if process.poll() is None:
-                process.terminate()
+                # Signal the whole process group first (graceful).
                 try:
-                    process.wait(timeout=2)
+                    os.killpg(process.pid, signal.SIGTERM)
+                except (ProcessLookupError, PermissionError):
+                    pass
+                try:
+                    process.wait(timeout=3)
                 except subprocess.TimeoutExpired:
-                    process.kill()
+                    # Forceful kill of the entire group.
+                    try:
+                        os.killpg(process.pid, signal.SIGKILL)
+                    except (ProcessLookupError, PermissionError):
+                        pass
+                    process.wait()
 
         elif backend == "winpty":
             pty_proc = session["pty_process"]

open_terminal/utils/runner.py

@@ -109,11 +109,21 @@ async def read_output(self, log_file) -> None:
     def write_input(self, data: bytes) -> None:
         os.write(self._master_fd, data)
 
+    def _signal_group(self, sig: int) -> None:
+        """Send *sig* to the child's entire process group.
+
+        Falls back to signalling just the leader if the group is already gone.
+        """
+        try:
+            os.killpg(self._process.pid, sig)
+        except (ProcessLookupError, PermissionError):
+            try:
+                self._process.send_signal(sig)
+            except ProcessLookupError:
+                pass
+
     def kill(self, force: bool = False) -> None:
-        if force:
-            self._process.kill()
-        else:
-            self._process.send_signal(signal.SIGTERM)
+        self._signal_group(signal.SIGKILL if force else signal.SIGTERM)
 
     async def wait(self) -> int:
         return await asyncio.to_thread(self._process.wait)
@@ -176,10 +186,12 @@ async def drain_input(self) -> None:
         await self._process.stdin.drain()
 
     def kill(self, force: bool = False) -> None:
-        if force:
-            self._process.kill()
-        else:
-            self._process.send_signal(signal.SIGTERM)
+        sig = signal.SIGKILL if force else signal.SIGTERM
+        try:
+            os.killpg(self._process.pid, sig)
+        except (ProcessLookupError, PermissionError, OSError):
+            # No dedicated process group — fall back to the child directly.
+            self._process.send_signal(sig)
 
     async def wait(self) -> int:
         await self._process.wait()

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "open-terminal"
-version = "0.11.9"
+version = "0.11.10"
 description = "A remote terminal API."
 readme = "README.md"
 authors = [