[GH-ISSUE #17782] issue: Web Search engine problem #18395

New Issue

2026-04-20T00:36:54-05:00

GiteaMirror commented

2026-04-20 00:36:54 -05:00

Originally created by @SaraDark on GitHub (Sep 26, 2025).
Original GitHub issue: https://github.com/open-webui/open-webui/issues/17782

Open WebUI Version

0.6.31

Operating System

Official Docker image on Linux Debian 12

Version: v0.6.31 and olders
Problem: When i search (searxng + firecrawl) i see boubled search, and not usable resoults like on screan below:

problem 2 is scraping only 1 page at time, i can by manuall test scraping 10 pagtes on one time.

my configuration:

Actual Behavior

Search doubled, 2 search not resoult, problem scrapping pages after searched.
Only 1 page at once scrapping is slowing qwery

Steps to Reproduce

Install from official docker image and configure like on screans

Originally created by @SaraDark on GitHub (Sep 26, 2025). Original GitHub issue: https://github.com/open-webui/open-webui/issues/17782 ### Open WebUI Version 0.6.31 ### Operating System Official Docker image on Linux Debian 12 Version: v0.6.31 and olders Problem: When i search (searxng + firecrawl) i see boubled search, and not usable resoults like on screan below: --------------- <img width="529" height="275" alt="Image" src="https://github.com/user-attachments/assets/e26d1a83-1ac5-4457-af76-06d4be09ac29" /> --------------- <img width="727" height="140" alt="Image" src="https://github.com/user-attachments/assets/171aae39-51c2-4e0c-97b3-5bdb867fc3f4" /> --------------- <img width="604" height="320" alt="Image" src="https://github.com/user-attachments/assets/199650a9-e012-4e05-8c7c-e3c0f3bd3b68" /> --------------- <img width="567" height="273" alt="Image" src="https://github.com/user-attachments/assets/469e7891-6342-4bd1-bf4c-06276a37d9b2" /> --------------- problem 2 is scraping only 1 page at time, i can by manuall test scraping 10 pagtes on one time. my configuration: <img width="2412" height="750" alt="Image" src="https://github.com/user-attachments/assets/ec5051a3-471e-487e-8d0f-d78cedca72ae" /> --------------- <img width="2401" height="815" alt="Image" src="https://github.com/user-attachments/assets/6d139deb-0b44-43f8-921a-83c8a84377f8" /> --------------- <img width="2420" height="1640" alt="Image" src="https://github.com/user-attachments/assets/04035c5a-9920-45af-984c-9e836a338bec" /> --------------- ### Actual Behavior Search doubled, 2 search not resoult, problem scrapping pages after searched. Only 1 page at once scrapping is slowing qwery ### Steps to Reproduce Install from official docker image and configure like on screans

GiteaMirror added the bug label 2026-04-20 00:36:54 -05:00

GiteaMirror closed this issue

2026-04-20 00:36:55 -05:00

GiteaMirror commented

2026-04-20 00:36:56 -05:00

@rgaricano commented on GitHub (Sep 26, 2025):

probably it only search 1 page for your searxng query,
are you tried just with: http://..../searxng/search?q=<query>

@rgaricano commented on GitHub (Sep 26, 2025): probably it only search 1 page for your searxng query, are you tried just with: `http://..../searxng/search?q=<query>`

GiteaMirror commented

2026-04-20 00:36:57 -05:00

@SaraDark commented on GitHub (Sep 26, 2025):

on older version (i dont remember what number version, i think v0.6.28) before refactoring its working without duplicated, but still with problem 1 query at one time problem, at current version i test config:

problem still exist with doubled searching (websearch not usable) and only 1 scrapp at one time

@SaraDark commented on GitHub (Sep 26, 2025): on older version (i dont remember what number version, i think v0.6.28) before refactoring its working without duplicated, but still with problem 1 query at one time problem, at current version i test config: <img width="817" height="168" alt="Image" src="https://github.com/user-attachments/assets/0afa802c-eea6-4935-98ec-2ea7975c3d6b" /> problem still exist with doubled searching (websearch not usable) and only 1 scrapp at one time

GiteaMirror commented

2026-04-20 00:36:57 -05:00

@rgaricano commented on GitHub (Sep 26, 2025):

there isn't 2 web search queries, only one, the other is the search queries generation (when is enabled).
This is the web_search flow:

graph TD
    UserMessage["User Message"] --> WebSearchCheck{{"ENABLE_WEB_SEARCH"}}
    
    WebSearchCheck -->|Disabled| NoWebSearch["Skip Web Search"]
    WebSearchCheck -->|Enabled| WebSearchHandler["chat_web_search_handler()"]
    
    WebSearchHandler --> StatusEmit1["Emit 'Searching the web' Status"]
    StatusEmit1 --> QueryGenCheck{{"ENABLE_SEARCH_QUERY_GENERATION"}}
    
    QueryGenCheck -->|Enabled| QueryGeneration["generate_queries()"]
    QueryGenCheck -->|Disabled| UseOriginal["Use Original Message"]
    
    QueryGeneration --> ParseQueries["Parse JSON Queries"]
    UseOriginal --> ProcessSearch["process_web_search()"]
    ParseQueries --> ProcessSearch
    
    ProcessSearch --> ParallelSearch["Parallel Search Execution"]
    ParallelSearch --> SearchEngines["18+ Search Engines"]
    
    SearchEngines --> URLCollection["URL Collection & Deduplication"]
    URLCollection --> WebLoaderCheck{{"BYPASS_WEB_SEARCH_WEB_LOADER"}}
    
    WebLoaderCheck -->|True| RawResults["Return Raw Search Results"]
    WebLoaderCheck -->|False| ContentLoading["Web Content Loading"]
    
    ContentLoading --> LoaderEngines["Web Loader Engines"]
    LoaderEngines --> EmbeddingCheck{{"BYPASS_WEB_SEARCH_EMBEDDING_AND_RETRIEVAL"}}
    
    EmbeddingCheck -->|True| FullContext["Inject Full Content as Context"]
    EmbeddingCheck -->|False| VectorStorage["Vector Database Storage"]
    
    RawResults --> FileAttachment["Attach to form_data.files"]
    FullContext --> FileAttachment
    VectorStorage --> FileAttachment
    
    FileAttachment --> StatusEmit2["Emit Search Complete Status"]
    StatusEmit2 --> ChatResponse["Enhanced Chat Response"]
    NoWebSearch --> ChatResponse

@rgaricano commented on GitHub (Sep 26, 2025): there isn't 2 web search queries, only one, the other is the search queries generation (when is enabled). This is the web_search flow: ```mermaid graph TD UserMessage["User Message"] --> WebSearchCheck{{"ENABLE_WEB_SEARCH"}} WebSearchCheck -->|Disabled| NoWebSearch["Skip Web Search"] WebSearchCheck -->|Enabled| WebSearchHandler["chat_web_search_handler()"] WebSearchHandler --> StatusEmit1["Emit 'Searching the web' Status"] StatusEmit1 --> QueryGenCheck{{"ENABLE_SEARCH_QUERY_GENERATION"}} QueryGenCheck -->|Enabled| QueryGeneration["generate_queries()"] QueryGenCheck -->|Disabled| UseOriginal["Use Original Message"] QueryGeneration --> ParseQueries["Parse JSON Queries"] UseOriginal --> ProcessSearch["process_web_search()"] ParseQueries --> ProcessSearch ProcessSearch --> ParallelSearch["Parallel Search Execution"] ParallelSearch --> SearchEngines["18+ Search Engines"] SearchEngines --> URLCollection["URL Collection & Deduplication"] URLCollection --> WebLoaderCheck{{"BYPASS_WEB_SEARCH_WEB_LOADER"}} WebLoaderCheck -->|True| RawResults["Return Raw Search Results"] WebLoaderCheck -->|False| ContentLoading["Web Content Loading"] ContentLoading --> LoaderEngines["Web Loader Engines"] LoaderEngines --> EmbeddingCheck{{"BYPASS_WEB_SEARCH_EMBEDDING_AND_RETRIEVAL"}} EmbeddingCheck -->|True| FullContext["Inject Full Content as Context"] EmbeddingCheck -->|False| VectorStorage["Vector Database Storage"] RawResults --> FileAttachment["Attach to form_data.files"] FullContext --> FileAttachment VectorStorage --> FileAttachment FileAttachment --> StatusEmit2["Emit Search Complete Status"] StatusEmit2 --> ChatResponse["Enhanced Chat Response"] NoWebSearch --> ChatResponse ```