[pull] master from AstrBotDevs:master

astrbot/core/computer/booters/base.py

@@ -36,7 +36,15 @@ def gui(self) -> GUIComponent | None:
 
     async def boot(self, session_id: str) -> None: ...
 
-    async def shutdown(self) -> None: ...
+    async def shutdown(self, **kwargs) -> None:
+        """Shut down the computer sandbox.
+
+        Subclasses may accept extra keyword arguments for
+        type-specific cleanup (e.g. ``delete_sandbox`` for
+        ShipyardNeoBooter).  The default implementation ignores
+        them.
+        """
+        ...
 
     async def upload_file(self, path: str, file_name: str) -> dict:
         """Upload file to the computer.

astrbot/core/computer/booters/shipyard_neo.py

@@ -1,5 +1,6 @@
 from __future__ import annotations
 
+import asyncio
 import os
 import shlex
 from typing import Any, cast
@@ -438,6 +439,9 @@ async def boot(self, session_id: str) -> None:
             ttl=self._ttl,
         )
 
+        # --- Readiness gate: wait until sandbox session is READY ---
+        await self._wait_until_ready(self._sandbox)
+
         self._shell = NeoShellComponent(self._sandbox)
         self._fs = NeoFileSystemComponent(self._sandbox, self._shell)
         self._python = NeoPythonComponent(self._sandbox)
@@ -455,6 +459,78 @@ async def boot(self, session_id: str) -> None:
             bool(self._bay_manager),
         )
 
+    async def _wait_until_ready(self, sandbox: Sandbox) -> None:
+        """Poll sandbox status until READY, or raise on FAILED / timeout.
+
+        Covers both warm-pool hits (near-instant) and cold starts (up to 180s).
+        On FAILED, EXPIRED, or timeout the sandbox is deleted before raising
+        so no orphan resources leak on Bay.
+        """
+        READINESS_TIMEOUT = 180  # seconds
+        POLL_INTERVAL = 2  # seconds
+
+        sandbox_id = sandbox.id
+        deadline = asyncio.get_running_loop().time() + READINESS_TIMEOUT
+
+        while True:
+            await sandbox.refresh()
+            status = getattr(sandbox.status, "value", str(sandbox.status))
+
+            if status == "ready":
+                logger.info(
+                    "[Computer] Sandbox %s is ready (profile=%s)",
+                    sandbox_id,
+                    sandbox.profile,
+                )
+                return
+
+            if status in {"failed", "expired"}:
+                logger.error(
+                    "[Computer] Sandbox %s reached terminal state: %s",
+                    sandbox_id,
+                    status,
+                )
+                try:
+                    await sandbox.delete()
+                except Exception as del_err:
+                    logger.warning(
+                        "[Computer] Failed to delete failed sandbox %s: %s",
+                        sandbox_id,
+                        del_err,
+                    )
+                raise RuntimeError(
+                    f"Sandbox {sandbox_id} is in terminal state: {status}"
+                )
+
+            remaining = deadline - asyncio.get_running_loop().time()
+            if remaining <= 0:
+                logger.error(
+                    "[Computer] Sandbox %s did not become ready within %ds "
+                    "(last status: %s)",
+                    sandbox_id,
+                    READINESS_TIMEOUT,
+                    status,
+                )
+                try:
+                    await sandbox.delete()
+                except Exception as del_err:
+                    logger.warning(
+                        "[Computer] Failed to delete timed-out sandbox %s: %s",
+                        sandbox_id,
+                        del_err,
+                    )
+                raise TimeoutError(
+                    f"Sandbox {sandbox_id} did not become ready within "
+                    f"{READINESS_TIMEOUT}s (last status: {status})"
+                )
+
+            logger.debug(
+                "[Computer] Sandbox %s status=%s, waiting...",
+                sandbox_id,
+                status,
+            )
+            await asyncio.sleep(POLL_INTERVAL)
+
     async def _resolve_profile(self, client: Any) -> str:
         """Pick the best profile for this session.
 
@@ -510,16 +586,41 @@ def _score(p: Any) -> tuple[int, int]:
 
         return chosen
 
-    async def shutdown(self) -> None:
+    async def shutdown(self, *, delete_sandbox: bool = False) -> None:
         if self._client is not None:
             sandbox_id = getattr(self._sandbox, "id", "unknown")
+
+            # Delete sandbox on Bay BEFORE closing the HTTP client.
+            # This is critical for cleanup — calling delete after
+            # __aexit__ would fail because the httpx session is already
+            # torn down.
+            if delete_sandbox and self._sandbox is not None:
+                try:
+                    logger.info(
+                        "[Computer] Deleting Shipyard Neo sandbox: id=%s", sandbox_id
+                    )
+                    await self._sandbox.delete()
+                    logger.info(
+                        "[Computer] Shipyard Neo sandbox deleted: id=%s", sandbox_id
+                    )
+                except Exception as e:
+                    logger.warning(
+                        "[Computer] Failed to delete sandbox %s (may already be "
+                        "cleaned up by Bay GC): %s",
+                        sandbox_id,
+                        e,
+                    )
+
             logger.info(
-                "[Computer] Shutting down Shipyard Neo sandbox: id=%s", sandbox_id
+                "[Computer] Shutting down Shipyard Neo sandbox client: id=%s",
+                sandbox_id,
             )
             await self._client.__aexit__(None, None, None)
             self._client = None
             self._sandbox = None
-            logger.info("[Computer] Shipyard Neo sandbox shut down: id=%s", sandbox_id)
+            logger.info(
+                "[Computer] Shipyard Neo sandbox client shut down: id=%s", sandbox_id
+            )
 
         # NOTE: We intentionally do NOT stop the Bay container here.
         # It stays running for reuse by future sessions.  The user can

astrbot/core/computer/computer_client.py

@@ -445,7 +445,22 @@ async def get_booter(
     if session_id in session_booter:
         booter = session_booter[session_id]
         if not await booter.available():
-            # rebuild
+            # Clean up old booter before rebuilding so sandbox resources
+            # on Bay (containers, volumes, networks) are not leaked.
+            # Only ShipyardNeoBooter supports delete_sandbox; other booters
+            # (local, boxlite, cua, etc.) are not backed by a remote sandbox
+            # manager and don't need it.
+            try:
+                if booter_type == "shipyard_neo":
+                    await booter.shutdown(delete_sandbox=True)
+                else:
+                    await booter.shutdown()
+            except Exception as shutdown_err:
+                logger.warning(
+                    "[Computer] Error shutting down stale booter for session %s: %s",
+                    session_id,
+                    shutdown_err,
+                )
             session_booter.pop(session_id, None)
     if session_id not in session_booter:
         uuid_str = uuid.uuid5(uuid.NAMESPACE_DNS, session_id).hex
@@ -509,7 +524,10 @@ async def get_booter(
         except Exception as e:
             logger.error(f"Error booting sandbox for session {session_id}: {e}")
             try:
-                await client.shutdown()
+                if booter_type == "shipyard_neo":
+                    await client.shutdown(delete_sandbox=True)
+                else:
+                    await client.shutdown()
             except Exception as shutdown_error:
                 logger.warning(
                     "Failed to shutdown sandbox after boot error for session %s: %s",

astrbot/core/core_lifecycle.py

@@ -59,6 +59,7 @@ def __init__(self, log_broker: LogBroker, db: BaseDatabase) -> None:
         self.subagent_orchestrator: SubAgentOrchestrator | None = None
         self.cron_manager: CronJobManager | None = None
         self.temp_dir_cleaner: TempDirCleaner | None = None
+        self._default_chat_provider_warning_emitted = False
 
         # 设置代理
         proxy_config = self.astrbot_config.get("http_proxy", "")
@@ -97,6 +98,47 @@ async def _init_or_reload_subagent_orchestrator(self) -> None:
         except Exception as e:
             logger.error(f"Subagent orchestrator init failed: {e}", exc_info=True)
 
+    def _warn_about_unset_default_chat_provider(self) -> None:
+        if self._default_chat_provider_warning_emitted:
+            return
+
+        pm = getattr(self, "provider_manager", None)
+        if not pm:
+            return
+
+        providers = pm.provider_insts
+        if len(providers) == 0:
+            return
+
+        provider_settings = getattr(pm, "provider_settings", None) or {}
+        default_id = provider_settings.get("default_provider_id")
+        fallback = pm.curr_provider_inst or providers[0]
+        fallback_id = fallback.provider_config.get("id") or "unknown"
+
+        if not default_id:
+            if len(providers) <= 1:
+                return
+            self._default_chat_provider_warning_emitted = True
+            logger.warning(
+                "Detected %d enabled chat providers but `provider_settings.default_provider_id` is empty. "
+                "AstrBot will use `%s` as the startup fallback chat provider. "
+                "Set a default chat model in the WebUI configuration page to avoid unexpected provider switching.",
+                len(providers),
+                fallback_id,
+            )
+            return
+
+        found = any((p.provider_config.get("id") == default_id) for p in providers)
+        if not found:
+            self._default_chat_provider_warning_emitted = True
+            logger.warning(
+                "Configured `default_provider_id` is `%s` but no enabled provider matches that ID. "
+                "AstrBot will use `%s` as the fallback chat provider. "
+                "Please check the WebUI configuration page.",
+                default_id,
+                fallback_id,
+            )
+
     async def initialize(self) -> None:
         """初始化 AstrBot 核心生命周期管理类.
 
@@ -201,7 +243,9 @@ async def initialize(self) -> None:
         await self.plugin_manager.reload()
 
         # 根据配置实例化各个 Provider
+        self._default_chat_provider_warning_emitted = False
         await self.provider_manager.initialize()
+        self._warn_about_unset_default_chat_provider()
 
         await self.kb_manager.initialize()