AdkApp - Session Service Builder with AlloyDB - Connection Error

[julienvincendeau-rc]

** Please make sure you read the contribution guide and file the issues in the right place. **
Contribution guide.

Describe the bug
I want to use DatabaseSessionService with my AlloyDB.
When I want to access on it, no error.
When I want to use with AdkApp, I've got an error.

To Reproduce
Here my code :

...

import logging
import sqlalchemy
import pg8000
from google.adk.sessions import DatabaseSessionService

....
# Initialize Vertex AI SDK
vertexai.init(
    project=GOOGLE_CLOUD_PROJECT,
    location=GOOGLE_CLOUD_LOCATION,
    staging_bucket=STAGING_BUCKET,
)
session_service = DatabaseSessionService(db_url=f"postgresql+pg8000://{db_user}:{db_pass}@{alloydb_instance_uri}/{db_name}", echo=True)


logger.info("deploying app...")
app = AdkApp(
    agent=root_agent,
    session_service_builder= lambda: DatabaseSessionService(
        db_url=f"postgresql+pg8000://{db_user}:{db_pass}@{alloydb_instance_uri}/{db_name}"
    ),
    enable_tracing=True,
)

Expected behavior
on this parts

session_service = DatabaseSessionService(db_url=f"postgresql+pg8000://{db_user}:{db_pass}@{alloydb_instance_uri}/{db_name}", echo=True)

I've got a sucess output

DEBUG:tzlocal:/etc/localtime found
DEBUG:tzlocal:2 found:
 {'/etc/timezone': 'Etc/UTC', '/etc/localtime is a symlink to': 'Etc/UTC'}
INFO:google_adk.google.adk.sessions.database_session_service:Local timezone: Etc/UTC
2025-10-27 18:05:37,115 INFO sqlalchemy.engine.Engine select pg_catalog.version()
INFO:sqlalchemy.engine.Engine:select pg_catalog.version()
2025-10-27 18:05:37,115 INFO sqlalchemy.engine.Engine [raw sql] ()
INFO:sqlalchemy.engine.Engine:[raw sql] ()
2025-10-27 18:05:37,371 INFO sqlalchemy.engine.Engine select current_schema()
INFO:sqlalchemy.engine.Engine:select current_schema()
2025-10-27 18:05:37,371 INFO sqlalchemy.engine.Engine [raw sql] ()
INFO:sqlalchemy.engine.Engine:[raw sql] ()
(...)

on this parts I've got an error

Traceback (most recent call last):
  File "/home/user/workspace.gcp/data_ai_agent_adk/global/deployment/deploy_agent.py", line 78, in <module>
    remote_app = agent_engines.update(
                 ^^^^^^^^^^^^^^^^^^^^^
  File "/home/user/workspace.gcp/data_ai_agent_adk/global/my_companion/.venv/lib/python3.12/site-packages/vertexai/agent_engines/__init__.py", line 383, in update
    return agent.update(
           ^^^^^^^^^^^^^
  File "/home/user/workspace.gcp/data_ai_agent_adk/global/my_companion/.venv/lib/python3.12/site-packages/vertexai/agent_engines/_agent_engines.py", line 836, in update
    created_resource = operation_future.result()
                       ^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/user/workspace.gcp/data_ai_agent_adk/global/my_companion/.venv/lib/python3.12/site-packages/google/api_core/future/polling.py", line 261, in result
    raise self._exception
google.api_core.exceptions.InvalidArgument: 400 The Reasoning Engine failed to be updated.
 Please refer to our documentation (https://cloud.google.com/vertex-ai/generative-ai/docs/agent-engine/troubleshooting/deploy) for checking logs and other troubleshooting tips. 3: The Reasoning Engine failed to be updated.

In the project where the agent is deployed, I've got more errors:

Desktop (please complete the following information):

• OS: [e.g. macOS, Linux, Windows] : GCP Workstation
• Python version(python -V):Python 3.12.3
• ADK version(pip show google-adk):1.17.0

Model Information:

• Are you using LiteLLM: Yes/No : No
• Which model is being used(e.g. gemini-2.5-pro) Gemini-2.5-Flash

Additional context
N/A

[julienvincendeau-rc]

Sorry, I found my first error, but I have a new one :(

remote_app = agent_engines.update(
        agent_engine=app,
        resource_name=os.getenv("AGENT_ENGINE_ID"),
        requirements=[
            'google-cloud-aiplatform[adk,agent-engines,reasoning-engines]', 
            'google-adk', 
            'llama-index', 
            'vertexai',
            'Pillow', 
            'python-dotenv', 
            'tqdm', 
            'requests', 
            'pydantic', 
            'cloudpickle',
            'sqlalchemy',
            'google-cloud-alloydb-connector[pg8000]'
        ],
        extra_packages=["./my_agent"],
        display_name=AGENT_DISPLAY_NAME
    )

I forgot to include pg800 in the requirements when creating the agent.
Well, it's done now, but I have a DB-side error now.`

[julienvincendeau-rc]

I found the resolution (thanks a lot @Matthieu68857)
The problem is not for ADK but for Vertex. During the engine creation the AlloyDB is not accessible : We need to create a PSC for the agent and configure it according this documentation https://cloud.google.com/agent-builder/agent-engine/deploy#psc-i

remote_agent = client.agent_engines.create(
    agent=local_agent,
    config={
        "psc_interface_config": {
            "network_attachment": "NETWORK_ATTACHMENT",
            "dns_peering_configs": [
                {
                    "domain": "DOMAIN_SUFFIX",
                    "target_project": "TARGET_PROJECT",
                    "target_network": "TARGET_NETWORK",
                },
            ],
        },
    },
)