[GH-ISSUE #12094] issue: Unable to Perform Vector Search Despite Having EMBEDDING Vectors in the Database #55135

Closed
opened 2026-05-05 17:11:52 -05:00 by GiteaMirror · 0 comments
Owner

Originally created by @chalitbkb on GitHub (Mar 26, 2025).
Original GitHub issue: https://github.com/open-webui/open-webui/issues/12094

Encountered two main problems
• After thoroughly configuring all settings and using the OpenAI API for embeddings to successfully upload the data file, I attempted to initiate a RAG query by asking a question. However, I discovered that the system was unable to search the VECTOR_DB data, even though the data exists in the database.
• Additionally, I encountered a problem uploading files: while the file named "heroui.md" successfully uploaded and began the embeddings process, the file named "nextdata.md" failed to upload.

Check Existing Issues

  • I have searched the existing issues and discussions.
  • I am using the latest version of Open WebUI.

Installation Method

Docker

Open WebUI Version

v0.5.20

Ollama Version (if applicable)

No response

Operating System

Windows 11

Browser (if applicable)

Chrome 134.0.6998.118

Confirmation

  • I have read and followed all instructions in README.md.
  • I am using the latest version of both Open WebUI and Ollama.
  • I have included the browser console logs.
  • I have included the Docker container logs.
  • I have listed steps to reproduce the bug in detail.

Expected Behavior

• When users upload data files via the OpenAI API for embeddings and initiate a RAG query, the system should be able to:
1. Perform the embeddings process successfully for all files.
2. Search the VECTOR_DB for relevant information based on the uploaded embedding vectors without any errors.

Actual Behavior

• Despite having successfully uploaded one data file and confirmed the presence of embedding vectors, the system failed to retrieve any information from the VECTOR_DB for that file.
• The file named "heroui.md" was able to upload and proceed with embeddings, but another file named "nextdata.md" failed to upload at all.

Steps to Reproduce

  1. Configure all necessary settings and connect to the VECTOR_DB.
  2. Upload a data file (e.g., "heroui.md") using the OpenAI API for embeddings and confirm it begins the embeddings process.
  3. Attempt to upload another data file (e.g., "nextdata.md") to observe if the upload fails.
  4. Verify that the embedding vectors for any successfully uploaded file have been created and stored in the database.
  5. Attempt to perform a RAG query related to the uploaded data.
  6. Observe that no data is returned for certain files, even though embeddings appear to be present, and some files fail to upload entirely.

Logs & Screenshots

Below are sample files where I’m experiencing this issue. Please note that "heroui.md" relates to the VECTOR_DB retrieval problem, whereas "nextdata.md" pertains to the unsuccessful upload issue. Even after changing the extension to .txt, the problem persisted.
heroui.md

nextdata.md

  • Successfully uploaded Knowledge and embeddings
    Image

Image

  • It can be seen that Knowledge has been selected from the Collection
    Image

  • Begin the process of testing by asking related questions. Notice that the data retrieval process is operating normally, indicating that VECTOR_DB data search has been enabled.
    Image

  • These are the results where relevant data could not be found even though data exists in the database. I tried several times but still no success.
    Image

Additional Information

I tried adding data myself, and was sometimes able to perform vector searches normally.

Image

Image

Originally created by @chalitbkb on GitHub (Mar 26, 2025). Original GitHub issue: https://github.com/open-webui/open-webui/issues/12094 Encountered two main problems • After thoroughly configuring all settings and using the OpenAI API for embeddings to successfully upload the data file, I attempted to initiate a RAG query by asking a question. However, I discovered that the system was unable to search the VECTOR_DB data, even though the data exists in the database. • Additionally, I encountered a problem uploading files: while the file named "heroui.md" successfully uploaded and began the embeddings process, the file named "nextdata.md" failed to upload. ### Check Existing Issues - [x] I have searched the existing issues and discussions. - [x] I am using the latest version of Open WebUI. ### Installation Method Docker ### Open WebUI Version v0.5.20 ### Ollama Version (if applicable) _No response_ ### Operating System Windows 11 ### Browser (if applicable) Chrome 134.0.6998.118 ### Confirmation - [x] I have read and followed all instructions in `README.md`. - [x] I am using the latest version of **both** Open WebUI and Ollama. - [x] I have included the browser console logs. - [x] I have included the Docker container logs. - [x] I have listed steps to reproduce the bug in detail. ### Expected Behavior • When users upload data files via the OpenAI API for embeddings and initiate a RAG query, the system should be able to: 1. Perform the embeddings process successfully for all files. 2. Search the VECTOR_DB for relevant information based on the uploaded embedding vectors without any errors. ### Actual Behavior • Despite having successfully uploaded one data file and confirmed the presence of embedding vectors, the system failed to retrieve any information from the VECTOR_DB for that file. • The file named "heroui.md" was able to upload and proceed with embeddings, but another file named "nextdata.md" failed to upload at all. ### Steps to Reproduce 1) Configure all necessary settings and connect to the VECTOR_DB. 2) Upload a data file (e.g., "heroui.md") using the OpenAI API for embeddings and confirm it begins the embeddings process. 3) Attempt to upload another data file (e.g., "nextdata.md") to observe if the upload fails. 4) Verify that the embedding vectors for any successfully uploaded file have been created and stored in the database. 5) Attempt to perform a RAG query related to the uploaded data. 6) Observe that no data is returned for certain files, even though embeddings appear to be present, and some files fail to upload entirely. ### Logs & Screenshots Below are sample files where I’m experiencing this issue. Please note that "heroui.md" relates to the VECTOR_DB retrieval problem, whereas "nextdata.md" pertains to the unsuccessful upload issue. Even after changing the extension to .txt, the problem persisted. [heroui.md](https://github.com/user-attachments/files/19468374/heroui.md) [nextdata.md](https://github.com/user-attachments/files/19468307/nextdata.md) - Successfully uploaded Knowledge and embeddings ![Image](https://github.com/user-attachments/assets/18744f1e-a33a-4fe9-bb5a-b7ba70895a74) ![Image](https://github.com/user-attachments/assets/ff08ea81-0f3e-4872-842d-8fbd00267837) - It can be seen that Knowledge has been selected from the Collection ![Image](https://github.com/user-attachments/assets/69f14d7f-79cb-411f-b7e9-75e38c316145) - Begin the process of testing by asking related questions. Notice that the data retrieval process is operating normally, indicating that VECTOR_DB data search has been enabled. ![Image](https://github.com/user-attachments/assets/3c300803-dba1-4ffd-839f-5ef8e2b710c5) - These are the results where relevant data could not be found even though data exists in the database. I tried several times but still no success. ![Image](https://github.com/user-attachments/assets/c1ba5004-03fa-46da-bc81-b0f0b89066c0) ### Additional Information I tried adding data myself, and was sometimes able to perform vector searches normally. ![Image](https://github.com/user-attachments/assets/0aabe504-b6e5-40de-b131-cb73be6492b1) ![Image](https://github.com/user-attachments/assets/e08fa71d-f1e9-49df-8689-d61d53cd7f18)
GiteaMirror added the bug label 2026-05-05 17:11:52 -05:00
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/open-webui#55135