Add debug logging for sandbox startup health checks (#12814)

Co-authored-by: openhands <openhands@all-hands.dev>

openhands/app_server/sandbox/docker_sandbox_service.py

@@ -230,7 +230,7 @@ class DockerSandboxService(SandboxService):
             except asyncio.CancelledError:
                 raise
             except Exception as exc:
-                # If the server is
+                # If the server has exceeded the startup grace period, it's an error
                 if sandbox_info.created_at < utc_now() - timedelta(
                     seconds=self.startup_grace_seconds
                 ):
@@ -239,6 +239,10 @@ class DockerSandboxService(SandboxService):
                     )
                     sandbox_info.status = SandboxStatus.ERROR
                 else:
+                    _logger.debug(
+                        f'Sandbox server not yet available (still starting): '
+                        f'{app_server_url} : {exc}'
+                    )
                     sandbox_info.status = SandboxStatus.STARTING
                 sandbox_info.exposed_urls = None
                 sandbox_info.session_api_key = None

openhands/app_server/sandbox/sandbox_service.py

@@ -1,4 +1,5 @@
 import asyncio
+import logging
 import time
 from abc import ABC, abstractmethod
 
@@ -18,6 +19,8 @@ from openhands.app_server.utils.docker_utils import (
 from openhands.sdk.utils.models import DiscriminatedUnionMixin
 from openhands.sdk.utils.paging import page_iterator
 
+_logger = logging.getLogger(__name__)
+
 SESSION_API_KEY_VARIABLE = 'OH_SESSION_API_KEYS_0'
 WEBHOOK_CALLBACK_VARIABLE = 'OH_WEBHOOKS_0_BASE_URL'
 ALLOW_CORS_ORIGINS_VARIABLE = 'OH_ALLOW_CORS_ORIGINS_0'
@@ -133,12 +136,17 @@ class SandboxService(ABC):
         Returns:
             True if agent server is alive, False otherwise
         """
+        url = None
         try:
             agent_server_url = self._get_agent_server_url(sandbox)
             url = f'{agent_server_url.rstrip("/")}/alive'
             response = await httpx_client.get(url, timeout=5.0)
             return response.is_success
-        except Exception:
+        except Exception as exc:
+            _logger.debug(
+                f'Agent server health check failed for sandbox {sandbox.id}'
+                f'{f" at {url}" if url else ""}: {exc}'
+            )
             return False
 
     def _get_agent_server_url(self, sandbox: SandboxInfo) -> str: