feat: multi-session terminal backend core (PR 1)

- Add TerminalSessionModel DB table with instance_id FK, name, status,
  created_at, last_activity_at, closed_at columns
- Add Alembic migration for terminal_sessions table
- Refactor TerminalManager to use composite key (instance_id, session_id)
  supporting up to 5 concurrent sessions per instance
- Add create_session, get_session, get_sessions_for_instance, close_session
- Preserve get_or_create_session for backward compatibility (default session)
- Fix attach_websocket to only close sockets within same session
- Add name (auto-generated 'Session N') and status tracking to TerminalSession
- Add 7 unit tests for multi-session logic

Quality gates: pytest (7 new passed, 174 total passed, 51 pre-existing failures)
This commit is contained in:
2026-05-28 11:38:22 +02:00
parent 22474cdba5
commit b55300ff6f
7 changed files with 652 additions and 71 deletions
+275 -46
View File
@@ -3,20 +3,37 @@
import asyncio
import logging
import uuid
from datetime import datetime, timezone
from fastapi import WebSocket
from src.database import SessionLocal
from src.models.terminal_session import TerminalSessionModel
from src.services.terminal_session import TerminalSession
logger = logging.getLogger(__name__)
class MaxSessionsExceededError(Exception):
"""Raised when the maximum number of terminal sessions per instance is reached."""
def __init__(self, instance_id: str, max_sessions: int = 5) -> None:
self.instance_id = instance_id
self.max_sessions = max_sessions
super().__init__(
f"Maximum of {max_sessions} terminal sessions reached for instance {instance_id}"
)
class TerminalManager:
"""Manages active terminal sessions with persistence support."""
# Maximum sessions per tool instance
MAX_SESSIONS_PER_INSTANCE = 5
def __init__(self) -> None:
# Track sessions by instance_id for persistence
self._sessions: dict[str, TerminalSession] = {}
# Track sessions by (instance_id, session_id) for multi-session support
self._sessions: dict[tuple[str, str], TerminalSession] = {}
self._idle_check_task: asyncio.Task | None = None
self._start_idle_check()
@@ -42,16 +59,131 @@ class TerminalManager:
async def _cleanup_idle_sessions(self) -> None:
"""Clean up sessions that have been idle for too long."""
idle_sessions = []
for instance_id, session in list(self._sessions.items()):
idle_keys = []
for (instance_id, session_id), session in list(self._sessions.items()):
if session.is_idle():
idle_sessions.append(instance_id)
for instance_id in idle_sessions:
logger.info("Cleaning up idle terminal session for instance %s", instance_id)
session = self._sessions.pop(instance_id, None)
idle_keys.append((instance_id, session_id))
for key in idle_keys:
instance_id, session_id = key
logger.info(
"Cleaning up idle terminal session %s for instance %s",
session_id,
instance_id,
)
session = self._sessions.pop(key, None)
if session:
await session.close()
# Update DB status fire-and-forget
asyncio.create_task(self._mark_closed_in_db(session_id))
async def _insert_db_session_row(
self,
session_id: str,
instance_id: uuid.UUID,
name: str,
) -> None:
"""Insert a TerminalSessionModel row into the database."""
try:
async with SessionLocal() as db_session:
db_row = TerminalSessionModel(
id=uuid.UUID(session_id),
instance_id=instance_id,
name=name,
status="active",
created_at=datetime.now(timezone.utc),
last_activity_at=datetime.now(timezone.utc),
)
db_session.add(db_row)
await db_session.commit()
logger.debug(
"Inserted terminal session row %s for instance %s",
session_id,
instance_id,
)
except Exception as exc:
logger.error("Failed to insert terminal session row: %s", exc)
async def _mark_closed_in_db(self, session_id: str) -> None:
"""Mark a terminal session as closed in the database."""
try:
async with SessionLocal() as db_session:
db_row = await db_session.get(
TerminalSessionModel, uuid.UUID(session_id)
)
if db_row:
db_row.status = "closed"
db_row.closed_at = datetime.now(timezone.utc)
await db_session.commit()
logger.debug(
"Marked terminal session %s as closed in DB", session_id
)
except Exception as exc:
logger.error("Failed to mark terminal session as closed in DB: %s", exc)
def _count_sessions_for_instance(self, instance_id_str: str) -> int:
"""Count active in-memory sessions for a given instance."""
return sum(1 for (iid, _sid) in self._sessions if iid == instance_id_str)
async def create_session(
self,
instance_id: uuid.UUID,
container_id: str,
startup_command: str | None = None,
name: str | None = None,
) -> TerminalSession:
"""Create a new terminal session for an instance.
Enforces a maximum of MAX_SESSIONS_PER_INSTANCE sessions per instance.
Inserts a DB row fire-and-forget.
Args:
instance_id: UUID of the tool instance.
container_id: Docker container ID.
startup_command: Optional startup command to run.
name: Optional session name (auto-generated if omitted).
Returns:
The newly created TerminalSession.
Raises:
MaxSessionsExceededError: If the instance already has max sessions.
"""
instance_id_str = str(instance_id)
if (
self._count_sessions_for_instance(instance_id_str)
>= self.MAX_SESSIONS_PER_INSTANCE
):
raise MaxSessionsExceededError(
instance_id_str, self.MAX_SESSIONS_PER_INSTANCE
)
session_id = str(uuid.uuid4())
session = TerminalSession(
session_id=session_id,
instance_id=instance_id,
container_id=container_id,
startup_command=startup_command,
name=name,
)
await session.start(startup_command=startup_command)
key = (instance_id_str, session_id)
self._sessions[key] = session
# Fire-and-forget DB insert
asyncio.create_task(
self._insert_db_session_row(session_id, instance_id, session.name)
)
logger.info(
"Created terminal session %s for instance %s (name=%s)",
session_id,
instance_id,
session.name,
)
return session
async def get_or_create_session(
self,
@@ -59,61 +191,128 @@ class TerminalManager:
container_id: str,
startup_command: str | None = None,
) -> TerminalSession:
"""Get existing session or create a new one."""
"""Get existing session or create a new one.
Backward-compatible alias that uses 'default' as the session_id.
"""
# Ensure idle check is running (lazy start)
self._start_idle_check()
instance_id_str = str(instance_id)
# Check for existing session
if instance_id_str in self._sessions:
session = self._sessions[instance_id_str]
key = (instance_id_str, "default")
# Check for existing default session
if key in self._sessions:
session = self._sessions[key]
# Check if session is still alive
if session.is_alive():
logger.debug("Reattaching to existing terminal session for instance %s", instance_id)
logger.debug(
"Reattaching to existing terminal session for instance %s",
instance_id,
)
return session
else:
# Session died, clean it up
logger.debug("Existing session for instance %s is dead, cleaning up", instance_id)
logger.debug(
"Existing session for instance %s is dead, cleaning up",
instance_id,
)
await session.close()
del self._sessions[instance_id_str]
# Create new session
logger.info("Creating new terminal session for instance %s", instance_id)
del self._sessions[key]
# Create new default session
logger.info(
"Creating new default terminal session for instance %s", instance_id
)
session_id = str(uuid.uuid4())
session = TerminalSession(session_id, instance_id, container_id, startup_command=startup_command)
session = TerminalSession(
session_id=session_id,
instance_id=instance_id,
container_id=container_id,
startup_command=startup_command,
name="Session 1",
)
await session.start(startup_command=startup_command)
self._sessions[instance_id_str] = session
self._sessions[key] = session
# Fire-and-forget DB insert
asyncio.create_task(
self._insert_db_session_row(session_id, instance_id, session.name)
)
return session
def get_session(
self,
instance_id: str,
session_id: str,
) -> TerminalSession | None:
"""Lookup a session by composite key."""
return self._sessions.get((instance_id, session_id))
def get_sessions_for_instance(
self,
instance_id: str,
) -> list[TerminalSession]:
"""Return all in-memory sessions for a given instance."""
return [
session
for (iid, _sid), session in self._sessions.items()
if iid == instance_id
]
async def close_session(
self,
instance_id: str,
session_id: str,
) -> None:
"""Close a specific session and update its DB status."""
key = (instance_id, session_id)
session = self._sessions.pop(key, None)
if session:
await session.close()
# Fire-and-forget DB update
asyncio.create_task(self._mark_closed_in_db(session_id))
logger.info(
"Closed terminal session %s for instance %s",
session_id,
instance_id,
)
async def attach_websocket(
self,
session: TerminalSession,
websocket: WebSocket,
) -> None:
"""Attach a WebSocket to an existing session."""
# Handle concurrent connections - close existing ones
"""Attach a WebSocket to an existing session.
Closes existing WebSocket connections only for this specific session.
"""
# Handle concurrent connections - close existing ones within the same session
if session.has_websockets():
logger.debug("Closing existing WebSocket connections for instance %s", session.instance_id)
logger.debug(
"Closing existing WebSocket connections for session %s (instance %s)",
session.session_id,
session.instance_id,
)
for ws in list(session._websockets):
try:
await ws.close(code=4000, reason="New connection established")
except Exception:
pass
pass # noqa: S110
session._websockets.clear()
# Attach new WebSocket
session.attach_websocket(websocket)
# Replay buffer
buffer = session.get_buffer()
if buffer:
try:
await websocket.send_bytes(buffer)
except Exception:
pass
pass # noqa: S110
async def detach_websocket(
self,
@@ -128,23 +327,53 @@ class TerminalManager:
instance_id: uuid.UUID,
container_id: str,
startup_command: str | None = None,
session_id: str | None = None,
) -> TerminalSession:
"""Reset a session by killing it and creating a new one."""
"""Reset a session by killing it and creating a new one.
Args:
instance_id: UUID of the tool instance.
container_id: Docker container ID.
startup_command: Optional startup command.
session_id: Specific session to reset. If None, resets the default session.
Returns:
The newly created TerminalSession.
"""
instance_id_str = str(instance_id)
target_session_id = session_id or "default"
key = (instance_id_str, target_session_id)
# Close existing session if any
if instance_id_str in self._sessions:
logger.debug("Resetting terminal session for instance %s", instance_id)
old_session = self._sessions.pop(instance_id_str)
if key in self._sessions:
logger.debug(
"Resetting terminal session %s for instance %s",
target_session_id,
instance_id,
)
old_session = self._sessions.pop(key)
await old_session.close()
# Create new session
session_id = str(uuid.uuid4())
session = TerminalSession(session_id, instance_id, container_id, startup_command=startup_command)
await session.start(startup_command=startup_command)
self._sessions[instance_id_str] = session
return session
# Fire-and-forget DB update for old session
asyncio.create_task(self._mark_closed_in_db(old_session.session_id))
# Create new session preserving the same session_id slot
new_session_id = str(uuid.uuid4())
new_session = TerminalSession(
session_id=new_session_id,
instance_id=instance_id,
container_id=container_id,
startup_command=startup_command,
name="Session 1" if target_session_id == "default" else None,
)
await new_session.start(startup_command=startup_command)
self._sessions[key] = new_session
# Fire-and-forget DB insert
asyncio.create_task(
self._insert_db_session_row(new_session_id, instance_id, new_session.name)
)
return new_session
async def close_all(self) -> None:
"""Close all active sessions."""
@@ -152,7 +381,7 @@ class TerminalManager:
self._sessions.clear()
for session in sessions:
await session.close()
if self._idle_check_task and not self._idle_check_task.done():
self._idle_check_task.cancel()
+55 -23
View File
@@ -18,18 +18,28 @@ logger = logging.getLogger(__name__)
class TerminalSession:
"""Manages a single terminal session connected to a docker container.
Supports persistent sessions that survive WebSocket disconnections.
Multiple WebSocket connections can attach/detach from the same session.
"""
# Circular buffer size (10KB)
BUFFER_SIZE = 10 * 1024
# Idle timeout in seconds (30 minutes)
IDLE_TIMEOUT = 30 * 60
def __init__(self, session_id: str, instance_id: uuid.UUID, container_id: str, startup_command: str | None = None) -> None:
# Session number counter per instance_id for auto-naming
_instance_counters: dict[str, int] = {}
def __init__(
self,
session_id: str,
instance_id: uuid.UUID,
container_id: str,
startup_command: str | None = None,
name: str | None = None,
) -> None:
self.session_id = session_id
self.instance_id = instance_id
self.container_id = container_id
@@ -38,37 +48,52 @@ class TerminalSession:
self._closed = False
self._master_fd: int | None = None
self._slave_fd: int | None = None
# Circular buffer for output replay
self._output_buffer: deque[bytes] = deque(maxlen=self.BUFFER_SIZE)
self._buffer_size = 0
# WebSocket connections
self._websockets: set[Any] = set()
# Activity tracking
self.last_activity = time.time()
# Terminal size
self._cols = 80
self._rows = 24
# Session metadata
self.name = name or self._generate_name(str(instance_id))
self.status: str = "active"
@classmethod
def _generate_name(cls, instance_id: str) -> str:
"""Generate an auto-incremented session name for the instance."""
count = cls._instance_counters.get(instance_id, 0) + 1
cls._instance_counters[instance_id] = count
return f"Session {count}"
async def start(self, startup_command: str | None = None) -> None:
"""Start the docker exec process with a shell using a PTY."""
# Create a pseudo-terminal on the host
self._master_fd, self._slave_fd = pty.openpty()
# Set the terminal size initially
self._set_terminal_size(self._cols, self._rows)
logger.debug(f"Starting terminal session {self.session_id} for container {self.container_id} with initial size {self._cols}x{self._rows}")
logger.debug(
f"Starting terminal session {self.session_id} for container {self.container_id} with initial size {self._cols}x{self._rows}"
)
# Build the shell command
if startup_command:
shell_cmd = f'bash -c "{startup_command}" || true; exec bash -il'
logger.debug(f"Using startup command for session {self.session_id}: {startup_command}")
logger.debug(
f"Using startup command for session {self.session_id}: {startup_command}"
)
else:
shell_cmd = "bash -il"
# Start docker exec with the slave fd as stdin/stdout/stderr
# Using -it because the slave fd IS a TTY
self.process = await asyncio.create_subprocess_exec(
@@ -85,11 +110,11 @@ class TerminalSession:
stdout=self._slave_fd,
stderr=self._slave_fd,
)
# Close slave fd in parent process
os.close(self._slave_fd)
self._slave_fd = None
self.last_activity = time.time()
def _set_terminal_size(self, cols: int, rows: int) -> None:
@@ -99,7 +124,7 @@ class TerminalSession:
return
# TIOCSWINSZ = 0x5414 on Linux
TIOCSWINSZ = 0x5414
size = struct.pack('HHHH', rows, cols, 0, 0)
size = struct.pack("HHHH", rows, cols, 0, 0)
try:
fcntl.ioctl(self._master_fd, TIOCSWINSZ, size)
logger.debug(f"Resized PTY to {cols}x{rows} (fd={self._master_fd})")
@@ -127,7 +152,7 @@ class TerminalSession:
"""Add data to circular buffer, maintaining size limit."""
self._output_buffer.append(data)
self._buffer_size += len(data)
# Trim if exceeds max size
while self._buffer_size > self.BUFFER_SIZE and self._output_buffer:
removed = self._output_buffer.popleft()
@@ -152,16 +177,16 @@ class TerminalSession:
if self._closed:
logger.warning("Cannot resize: session is closed")
return
# Only resize if dimensions actually changed
if cols == self._cols and rows == self._rows:
return
self._cols = cols
self._rows = rows
logger.debug(f"resize() called for session {self.session_id}: {cols}x{rows}")
self._set_terminal_size(cols, rows)
# Docker exec -it creates its own PTY inside the container,
# so host PTY resize doesn't propagate to the container shell.
# Send SIGWINCH to the docker exec process on the host.
@@ -170,14 +195,19 @@ class TerminalSession:
if self.process and self.process.pid:
try:
os.kill(self.process.pid, signal.SIGWINCH)
logger.debug(f"Sent SIGWINCH to docker exec process {self.process.pid} for session {self.session_id}")
logger.debug(
f"Sent SIGWINCH to docker exec process {self.process.pid} for session {self.session_id}"
)
except ProcessLookupError:
logger.warning(f"docker exec process {self.process.pid} not found for session {self.session_id}")
logger.warning(
f"docker exec process {self.process.pid} not found for session {self.session_id}"
)
except Exception as e:
logger.warning(f"Failed to send SIGWINCH: {e}")
async def reset(self) -> None:
"""Reset the session by killing the process and clearing state."""
self.status = "resetting"
await self.close()
self._closed = False
self._output_buffer.clear()
@@ -186,18 +216,20 @@ class TerminalSession:
self.process = None
self._master_fd = None
self._slave_fd = None
self.status = "active"
async def close(self) -> None:
"""Close the session and cleanup."""
if self._closed:
return
self._closed = True
self.status = "closed"
if self._master_fd is not None:
try:
os.close(self._master_fd)
except OSError:
pass
pass # noqa: S110
self._master_fd = None
if self.process is not None:
@@ -240,7 +272,7 @@ class TerminalSession:
await ws.send_bytes(data)
except Exception:
dead_sockets.add(ws)
# Clean up dead sockets
for ws in dead_sockets:
self._websockets.discard(ws)