[GH-ISSUE #24026] issue: Long MP3 transcription remains pending in single-replica Docker deployment (v0.9.1) #58825

New Issue

GiteaMirror · 2026-05-06T00:15:07-05:00

GiteaMirror commented

2026-05-06 00:15:07 -05:00

Originally created by @kaeferpsd on GitHub (Apr 23, 2026).
Original GitHub issue: https://github.com/open-webui/open-webui/issues/24026

Check Existing Issues

I have searched for any existing and/or related issues.
I have searched for any existing and/or related discussions.
I have also searched in the CLOSED issues AND CLOSED discussions and found no related items (your issue might already be addressed on the development branch!).
I am using the latest version of Open WebUI.

Installation Method

Git Clone

Open WebUI Version

0.9.1

Ollama Version (if applicable)

latest

Operating System

Windows 11 Enterprise

Browser (if applicable)

Chrome 147.0.7727.102

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 provided every relevant configuration, setting, and environment variable used in my setup.
I have clearly listed every relevant configuration, custom setting, environment variable, and command-line option that influences my setup (such as Docker Compose overrides, .env values, browser settings, authentication configurations, etc).
I have documented step-by-step reproduction instructions that are precise, sequential, and leave nothing to interpretation. My steps:
Start with the initial platform/version/OS and dependencies used,
Specify exact install/launch/configure commands,
List URLs visited, user input (incl. example values/emails/passwords if needed),
Describe all options and toggles enabled or changed,
Include any files or environmental changes,
Identify the expected and actual result at each stage,
Ensure any reasonably skilled user can follow and hit the same issue.

Expected Behavior

Expected Behavior In a single-replica Docker deployment, long MP3 uploads in Notes should always end in a terminal state:

completed with transcript content, or
failed with a clear error message.
After transcription, Notes Enhance should use the transcript and should not show fallback text.

Actual Behavior On v0.9.1 (single replica), long MP3 uploads start normally but remain pending indefinitely.

Observed log flow reaches:

transcribe
convert_audio_to_mp3
Processing audio with duration ...
Detected language ...
After that, no final status update is written (completed/failed), the file stays pending, and Notes Enhance returns fallback text like “Could not extract content / No usable context could be extracted.”

Regression note: This same workflow worked before on v0.8.11.

Actual Behavior

On v0.9.1 in a single-replica Docker setup, long MP3 uploads in Notes do not reach a terminal state.

What happens:

File upload succeeds (POST /api/v1/files returns 200).
Transcription starts and logs show:
- transcribe: ...
- convert_audio_to_mp3: ...
- Processing audio with duration ...
- Detected language ...
After that, there is no final status transition in the file row.
file.data.status remains pending indefinitely (no completed, no failed, no error text).
Notes Enhance then shows fallback output like “Could not extract content / No usable context could be extracted.”

Regression: the same workflow worked before in v0.8.11.

Steps to Reproduce

Deploy Open WebUI v0.9.1 with Docker Compose using one open-webui container (single replica), plus Postgres and Redis.
Log in to Open WebUI and open Notes.
Create a new note.
Upload a long MP3 file (reproducible with ~60+ minutes audio, around ~50–60 MB).
Wait for processing (10+ minutes).
Check logs on the open-webui container. You can see lines similar to:
- transcribe: ...
- convert_audio_to_mp3: ...
- Processing audio with duration ...
- Detected language ...
Check DB status for the latest file (example query):
SELECT id, filename, (data->>'status') AS status, data::text FROM file ORDER BY created_at DESC LIMIT 5;
Run Enhance in the note.

Observed result:

The uploaded file remains pending (no terminal state),
and the note returns fallback text (“Could not extract content”).

Logs & Screenshots

Single-replica NAS1 run (open-webui container, image ghcr.io/open-webui/open-webui:0.9.1) while uploading a large MP3 (~44 MB).

Runtime logs

2026-04-23 07:15:16.766 | INFO  | open_webui.routers.audio:transcribe:1102 - transcribe: /app/backend/data/uploads/40235237-7b62-407c-a79c-e133256551b3_2026-02-04 09-30-02.mp3 {'language': 'de'}
2026-04-23 07:15:55.205 | INFO  | open_webui.routers.audio:convert_audio_to_mp3:123 - Converted /app/backend/data/uploads/40235237-7b62-407c-a79c-e133256551b3_2026-02-04 09-30-02.mp3 to /app/backend/data/uploads/40235237-7b62-407c-a79c-e133256551b3_2026-02-04 09-30-02.mp3
Chunk paths: ['/app/backend/data/uploads/40235237-7b62-407c-a79c-e133256551b3_2026-02-04 09-30-02_compressed.mp3']
2026-04-23 07:16:14.967 | INFO  | faster_whisper.transcribe:transcribe:881 - Processing audio with duration 32:20.281
2026-04-23 07:16:17.215 | INFO  | open_webui.routers.audio:transcription_handler:660 - Detected language 'de' with probability 1.000000

DB snapshot after waiting (still pending)

id:            40235237-7b62-407c-a79c-e133256551b3
filename:      2026-02-04 09-30-02.mp3
status:        pending
error:         (empty)
content_flag:  no_content
updated_utc:   2026-04-23 07:15:16

No transition to completed or failed is written for this file.

### Additional Information


Und für **Additional Information** kannst du 1 Zeile ergänzen:

```markdown
Regression note: this workflow worked before in v0.8.11 and started failing after upgrading to v0.9.1.

Originally created by @kaeferpsd on GitHub (Apr 23, 2026). Original GitHub issue: https://github.com/open-webui/open-webui/issues/24026 ### Check Existing Issues - [x] I have searched for any existing and/or related issues. - [x] I have searched for any existing and/or related discussions. - [x] I have also searched in the CLOSED issues AND CLOSED discussions and found no related items (your issue might already be addressed on the development branch!). - [x] I am using the latest version of Open WebUI. ### Installation Method Git Clone ### Open WebUI Version 0.9.1 ### Ollama Version (if applicable) latest ### Operating System Windows 11 Enterprise ### Browser (if applicable) Chrome 147.0.7727.102 ### 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 **provided every relevant configuration, setting, and environment variable used in my setup.** - [x] I have clearly **listed every relevant configuration, custom setting, environment variable, and command-line option that influences my setup** (such as Docker Compose overrides, .env values, browser settings, authentication configurations, etc). - [x] I have documented **step-by-step reproduction instructions that are precise, sequential, and leave nothing to interpretation**. My steps: - Start with the initial platform/version/OS and dependencies used, - Specify exact install/launch/configure commands, - List URLs visited, user input (incl. example values/emails/passwords if needed), - Describe all options and toggles enabled or changed, - Include any files or environmental changes, - Identify the expected and actual result at each stage, - Ensure any reasonably skilled user can follow and hit the same issue. ### Expected Behavior Expected Behavior In a single-replica Docker deployment, long MP3 uploads in Notes should always end in a terminal state: completed with transcript content, or failed with a clear error message. After transcription, Notes Enhance should use the transcript and should not show fallback text. Actual Behavior On v0.9.1 (single replica), long MP3 uploads start normally but remain pending indefinitely. Observed log flow reaches: transcribe convert_audio_to_mp3 Processing audio with duration ... Detected language ... After that, no final status update is written (completed/failed), the file stays pending, and Notes Enhance returns fallback text like “Could not extract content / No usable context could be extracted.” Regression note: This same workflow worked before on v0.8.11. ### Actual Behavior On `v0.9.1` in a **single-replica** Docker setup, long MP3 uploads in Notes do not reach a terminal state. What happens: - File upload succeeds (`POST /api/v1/files` returns 200). - Transcription starts and logs show: - `transcribe: ...` - `convert_audio_to_mp3: ...` - `Processing audio with duration ...` - `Detected language ...` - After that, there is no final status transition in the `file` row. - `file.data.status` remains `pending` indefinitely (no `completed`, no `failed`, no error text). - Notes Enhance then shows fallback output like “Could not extract content / No usable context could be extracted.” Regression: the same workflow worked before in `v0.8.11`. ### Steps to Reproduce 1. Deploy Open WebUI `v0.9.1` with Docker Compose using **one** `open-webui` container (single replica), plus Postgres and Redis. 2. Log in to Open WebUI and open **Notes**. 3. Create a new note. 4. Upload a long MP3 file (reproducible with ~60+ minutes audio, around ~50–60 MB). 5. Wait for processing (10+ minutes). 6. Check logs on the `open-webui` container. You can see lines similar to: - `transcribe: ...` - `convert_audio_to_mp3: ...` - `Processing audio with duration ...` - `Detected language ...` 7. Check DB status for the latest file (example query): `SELECT id, filename, (data->>'status') AS status, data::text FROM file ORDER BY created_at DESC LIMIT 5;` 8. Run **Enhance** in the note. Observed result: - The uploaded file remains `pending` (no terminal state), - and the note returns fallback text (“Could not extract content”). ### Logs & Screenshots Single-replica NAS1 run (`open-webui` container, image `ghcr.io/open-webui/open-webui:0.9.1`) while uploading a large MP3 (~44 MB). ### Runtime logs ```text 2026-04-23 07:15:16.766 | INFO | open_webui.routers.audio:transcribe:1102 - transcribe: /app/backend/data/uploads/40235237-7b62-407c-a79c-e133256551b3_2026-02-04 09-30-02.mp3 {'language': 'de'} 2026-04-23 07:15:55.205 | INFO | open_webui.routers.audio:convert_audio_to_mp3:123 - Converted /app/backend/data/uploads/40235237-7b62-407c-a79c-e133256551b3_2026-02-04 09-30-02.mp3 to /app/backend/data/uploads/40235237-7b62-407c-a79c-e133256551b3_2026-02-04 09-30-02.mp3 Chunk paths: ['/app/backend/data/uploads/40235237-7b62-407c-a79c-e133256551b3_2026-02-04 09-30-02_compressed.mp3'] 2026-04-23 07:16:14.967 | INFO | faster_whisper.transcribe:transcribe:881 - Processing audio with duration 32:20.281 2026-04-23 07:16:17.215 | INFO | open_webui.routers.audio:transcription_handler:660 - Detected language 'de' with probability 1.000000 DB snapshot after waiting (still pending) id: 40235237-7b62-407c-a79c-e133256551b3 filename: 2026-02-04 09-30-02.mp3 status: pending error: (empty) content_flag: no_content updated_utc: 2026-04-23 07:15:16 No transition to completed or failed is written for this file. ### Additional Information Und für **Additional Information** kannst du 1 Zeile ergänzen: ```markdown Regression note: this workflow worked before in v0.8.11 and started failing after upgrading to v0.9.1.

GiteaMirror added the bug label 2026-05-06 00:15:07 -05:00

Sign in to join this conversation.

Branches Tags

1 Participants

Notifications

Due Date

No due date set.

Dependencies

No dependencies set.

Reference: github-starred/open-webui#58825