Unable to use open-webui without setting up "Memory"

[Kisaragi-ng]

Bug Report

Description

Bug Summary:
In version v0.1.125, open-webui is not passing chat into ollama, causing the chat just hang / stuck. However, by enabling Memory and creating one personality make the chat works properly.

Steps to Reproduce:

1. using docker-compose.yml, run docker compose up
2. install phi3:latest from ollama library
3. begin chat using template
4. ollama doesn't respond to chat
5. enable Memory, create one personality, restart chat
6. chat works as intended

Expected Behavior:
open-webui should still works without setting up Memory, as this feature is Experimental (and from my understanding, this is not mandatory).

Actual Behavior:
open-webui doesn't respond to chat.

Environment

• Open WebUI Version: v0.1.125

• Ollama (if applicable): 0.1.38

• Operating System: docker

• Browser (if applicable): Firefox 115.11.0esr, Edge 125.0.2535.51 (Official build) (64-bit)

Reproduction Details

Confirmation:

• I have read and followed all the instructions provided in the README.md.
• I am on the latest version of both Open WebUI and Ollama.
• I have included the browser console logs.
• I have included the Docker container logs.

Logs and Screenshots

Browser Console Logs:

firefox:

Backend config: 
Object { status: true, name: "Open WebUI", version: "0.1.125", auth: true, default_locale: "en-US", images: false, default_models: null, default_prompt_suggestions: (6) […], trusted_header_auth: false, admin_export_enabled: true }
0.3690807d.js:1:37714
submitPrompt <empty string> 4.dc01e3b1.js:5:2577
modelId phi3:latest 4.dc01e3b1.js:5:3857
Uncaught (in promise) TypeError: l.userContext is null
    Immutable 38
    <anonymous> https://mydomain.com/:79
    promise callback* https://mydomain.com/:78
4.dc01e3b1.js:8:3
    we Immutable
    AsyncFunctionThrow self-hosted:857
    Immutable 17
    InterpretGeneratorResume self-hosted:1461
    AsyncFunctionNext self-hosted:852
    (Async: async)
    F Immutable
    map self-hosted:221
    Immutable 6
    InterpretGeneratorResume self-hosted:1461
    AsyncFunctionNext self-hosted:852
    (Async: async)
    F Immutable
    map self-hosted:221
    Immutable 6
    InterpretGeneratorResume self-hosted:1461
    AsyncFunctionNext self-hosted:852
    Immutable 3
    <anonymous> https://mydomain.com/:79
    (Async: promise callback)
    <anonymous> https://mydomain.com/:78

edge:

Backend config: 
Object { status: true, name: "Open WebUI", version: "0.1.125", auth: true, default_locale: "en-US", images: false, default_models: "phi3:latest", default_prompt_suggestions: (6) […], trusted_header_auth: false, admin_export_enabled: true }
0.3690807d.js:1:37714
setting default models globally index.4df4e131.js:93:23328
submitPrompt <empty string> 4.dc01e3b1.js:5:2577
modelId phi3:latest 4.dc01e3b1.js:5:3857
Uncaught (in promise) TypeError: l.userContext is null
    Immutable 5
4.dc01e3b1.js:8:3

Docker Container Logs:
unfortunately, nothing interesting:

open-webui  | INFO:     127.0.0.1:57778 - "GET /health HTTP/1.1" 200 OK
open-webui  | INFO:     127.0.0.1:55764 - "GET /health HTTP/1.1" 200 OK
open-webui  | INFO:     127.0.0.1:58104 - "GET /health HTTP/1.1" 200 OK
open-webui  | INFO:     127.0.0.1:59696 - "GET /health HTTP/1.1" 200 OK
open-webui  | INFO:     127.0.0.1:38172 - "GET /health HTTP/1.1" 200 OK
open-webui  | INFO:     ****:0 - "GET / HTTP/1.1" 200 OK
open-webui  | INFO:     ****:0 - "GET /api/config HTTP/1.1" 200 OK
open-webui  | INFO:     ****:0 - "GET /api/v1/auths/ HTTP/1.1" 200 OK
open-webui  | INFO:     ****:0 - "GET /api/changelog HTTP/1.1" 200 OK
open-webui  | INFO:apps.ollama.main:get_all_models()
open-webui  | INFO:apps.openai.main:get_all_models()
ollama      | [GIN] 2024/05/20 - 04:58:30 | 200 |     416.606µs |      172.21.0.3 | GET      "/api/tags"
open-webui  | INFO:     ****:0 - "GET /ollama/api/tags HTTP/1.1" 200 OK
open-webui  | INFO:apps.openai.main:get_all_models()
open-webui  | INFO:     ****:0 - "GET /litellm/api/v1/models HTTP/1.1" 200 OK
open-webui  | INFO:     ****:0 - "GET /openai/api/models HTTP/1.1" 200 OK
open-webui  | INFO:     ****:0 - "GET /api/v1/modelfiles/ HTTP/1.1" 200 OK
open-webui  | INFO:     ****:0 - "GET /api/v1/prompts/ HTTP/1.1" 200 OK
open-webui  | INFO:     ****:0 - "GET /api/v1/documents/ HTTP/1.1" 200 OK
open-webui  | INFO:     ****:0 - "GET /api/v1/chats/tags/all HTTP/1.1" 200 OK
open-webui  | INFO:apps.ollama.main:get_all_models()
ollama      | [GIN] 2024/05/20 - 04:58:31 | 200 |     393.705µs |      172.21.0.3 | GET      "/api/tags"
open-webui  | INFO:     ****:0 - "GET /ollama/api/tags HTTP/1.1" 200 OK
open-webui  | INFO:apps.openai.main:get_all_models()
open-webui  | INFO:apps.openai.main:get_all_models()
open-webui  | INFO:     ****:0 - "GET /litellm/api/v1/models HTTP/1.1" 200 OK
open-webui  | INFO:     ****:0 - "GET /openai/api/models HTTP/1.1" 200 OK
ollama      | [GIN] 2024/05/20 - 04:58:31 | 200 |        23.7µs |      172.21.0.3 | GET      "/api/version"
open-webui  | INFO:     ****:0 - "GET /api/v1/chats/ HTTP/1.1" 200 OK
open-webui  | INFO:     ****:0 - "GET /ollama/api/version HTTP/1.1" 200 OK
open-webui  | INFO:     ****:0 - "POST /api/v1/chats/new HTTP/1.1" 200 OK
open-webui  | INFO:     ****:0 - "GET /api/v1/chats/ HTTP/1.1" 200 OK
open-webui  | INFO:     127.0.0.1:39706 - "GET /health HTTP/1.1" 200 OK

Screenshots (if applicable):
Memory disabled:

Memory enabled:

Memory entry:

Installation Method

Docker version 26.1.2, build 211e74b

Additional Information

I'm willing to provide additional information if required. during this open-webui usage, I have already make sure there are no ad blocker running active, and I also have tested this in private window without addon enabled resulting the same result. I also found this error after updating my docker image from adb86c02cf4b to b0ef2a0e3744 (current main docker image).

[mptyl]

Same problem + error in connecting with openAI. An error windows pop-up appears with a join error. An OpenAI model can only be used by deleting the system prompt.

If you enable the memory, you can use the System prompt with OpenAI models like gpt-3.5.

To replicate the event:

1. remove the memory
2. write something in system prompt, whatever you want
3. use an OpenAI model (exception is thrown) or use a local model (hangs up)

[tjbck]

Please pull the current latest main, and let me know if the issue persists!

[CultusMechanicus]

It's still not working properly, now disabling the Memory toggle also disables the System Prompt too.

[tjbck]

@CultusMechanicus Added a fix with #2427, let me know if the issue persists!

[JamesBedwell]

This seems to be related to my issue #2441. I enabled memory and it allows me to send the message once again, but it ignores the system prompt. I have not yet tried the latest one with the fix #2427.