Vertex AI RAG Engine - "Failed to search for project" Error (HTTP 500) - Possible Service Outage

[Yoshinori]

I'm experiencing an issue with Vertex AI RAG Engine that appears to be a backend service problem rather than a code issue.

Problem Description

When using Vertex AI with RAG functionality, I'm consistently getting the following error:

google.api_core.exceptions.InternalServerError: 500 Failed to search for project {project ID}

Important Details

• The error started occurring this morning (around 8:47 on 2025 Mar 4 JST (23:47 on Mar 3 UTC))
• Everything was working fine until last night (23:36 on 2025 Mar 3  JST (14:36 on Mar 3 UTC))
• The error occurs both in my Python code AND in the Google Cloud Console when using RAG Engine
• When I use Generative AI without RAG functionality, it works fine
• I've confirmed the RAG corpus exists and is in ACTIVE state using list_corpora method
• Changing model, location or recreating corpus doesn't resolve the issue
• Vertex AI API is enabled (confirmed with gcloud services list)

Code

vertexai.init(project=PROJECT_ID, location=LOCATION)

rag_retrieval_config = rag.RagRetrievalConfig(
    top_k=50,
    filter=rag.Filter(
        vector_distance_threshold=0.5,
    ),
)

rag_retrieval_tool = Tool.from_retrieval(
    retrieval=rag.Retrieval(
        source=rag.VertexRagStore(
            rag_resources=[
                rag.RagResource(
                    rag_corpus="projects/PROJECT/locations/us-central1/ragCorpora/CORPUS_ID",
                )
            ],
            rag_retrieval_config=rag_retrieval_config,
        ),
    )
)

rag_model = GenerativeModel(
    model_name=MODEL_NAME,
    tools=[rag_retrieval_tool],
    system_instruction=INSTRUCTION,
)

response = rag_model.generate_content(
    "Hi are you awake?"
)

What I've Tried

• Verified IAM permissions (all service accounts have appropriate roles)
• Confirmed the RAG corpus exists and is active
• Tried different models and locations
• Verified API is enabled
• Recreated RAG corpus and re-imported files
• Tested in Google Cloud Console (same error occurs)

Since the error occurs in the Google Cloud Console as well, I believe this is a backend service issue rather than a problem with my code or configuration.

Is anyone else experiencing this issue? Is there a known outage or problem with the Vertex AI RAG service?

Any help or information would be greatly appreciated.

[jree]

I encountered the same issue. First occurrence at Mar 3, 2025 9:21:18 PM UTC

[Smart123s]

Facing the same issue.

[Yoshinori · Accepted solution]

It started working on my end! Seems that the issue is solved.