langgenius/dify
1
0
Fork 0
mirror of https://github.com/langgenius/dify.git synced 2026-07-21 03:55:25 -04:00
Code Issues 775 Packages Projects Releases 10 Wiki Activity

1.11.1 Knowledge Base Issues #21048

New Issue
Closed
opened 2026-02-21 20:10:25 -05:00 by yindo · 3 comments
yindo commented 2026-02-21 20:10:25 -05:00
Owner
Copy Link
Copy Source

Originally created by @froukee on GitHub (Dec 15, 2025).

Originally assigned to: @fatelei on GitHub.

Self Checks

  • I have read the Contributing Guide and Language Policy.
  • This is only for bug report, if you would like to ask a question, please head to Discussions.
  • I have searched for existing issues search for existing issues, including closed ones.
  • I confirm that I am using English to submit this report, otherwise it will be closed.
  • 【中文用户 & Non English User】请使用英语提交，否则会被关闭 ：）
  • Please do not modify this template :) and fill in all the required fields.

Dify version

1.11.1

Cloud or Self Hosted

Self Hosted (Docker)

Steps to reproduce

1.11.1 Knowledge Base Issues

  1. Manually added segments cannot be automatically indexed. Previously, in version 1.8.1, adding a segment allowed direct retrieval, but now it requires manually closing and reopening the knowledge base for retrieval.

  2. Metadata cannot use the value 0; if it's 0, it cannot be searched. Additionally, the format is incorrect. I selected "8080" for the number, but the search result is a string instead of "number." Similarly, another field, "number 3," yields a "number" result.

  3. Adding segments to Simple QA allows uploading images to the answer, but it cannot retrieve results. I understand that at least the uploaded image link as a string should work, right? For example, if my Q is "test image," and A is an uploaded image and some text, this QA cannot be retrieved regardless of where I test it. Strangely, when testing the knowledge retrieval function in the workflow, there are no results, but it appears when counting segments in the knowledge base.

Image Image Image

✔️ Expected Behavior

Normal use of recall function

❌ Actual Behavior

No response

Originally created by @froukee on GitHub (Dec 15, 2025). Originally assigned to: @fatelei on GitHub. ### Self Checks - [x] I have read the [Contributing Guide](https://github.com/langgenius/dify/blob/main/CONTRIBUTING.md) and [Language Policy](https://github.com/langgenius/dify/issues/1542). - [x] This is only for bug report, if you would like to ask a question, please head to [Discussions](https://github.com/langgenius/dify/discussions/categories/general). - [x] I have searched for existing issues [search for existing issues](https://github.com/langgenius/dify/issues), including closed ones. - [x] I confirm that I am using English to submit this report, otherwise it will be closed. - [x] 【中文用户 & Non English User】请使用英语提交，否则会被关闭 ：） - [x] Please do not modify this template :) and fill in all the required fields. ### Dify version 1.11.1 ### Cloud or Self Hosted Self Hosted (Docker) ### Steps to reproduce 1.11.1 Knowledge Base Issues 1. Manually added segments cannot be automatically indexed. Previously, in version 1.8.1, adding a segment allowed direct retrieval, but now it requires manually closing and reopening the knowledge base for retrieval. 2. Metadata cannot use the value 0; if it's 0, it cannot be searched. Additionally, the format is incorrect. I selected "8080" for the number, but the search result is a string instead of "number." Similarly, another field, "number 3," yields a "number" result. 3. Adding segments to Simple QA allows uploading images to the answer, but it cannot retrieve results. I understand that at least the uploaded image link as a string should work, right? For example, if my Q is "test image," and A is an uploaded image and some text, this QA cannot be retrieved regardless of where I test it. Strangely, when testing the knowledge retrieval function in the workflow, there are no results, but it appears when counting segments in the knowledge base. <img width="333" height="150" alt="Image" src="https://github.com/user-attachments/assets/93f81e1d-889a-422a-851a-d7446bf00475" /> <img width="231" height="108" alt="Image" src="https://github.com/user-attachments/assets/43431d9a-3425-4243-9054-c7b302e646f5" /> <img width="385" height="182" alt="Image" src="https://github.com/user-attachments/assets/543397f0-e312-4682-b531-31f4281e9b28" /> ### ✔️ Expected Behavior Normal use of recall function ### ❌ Actual Behavior _No response_
yindo added the 🐞 bug label 2026-02-21 20:10:25 -05:00
yindo closed this issue 2026-02-21 20:10:25 -05:00
yindo commented 2026-02-21 20:10:26 -05:00
Author
Owner
Copy Link
Copy Source

@dosubot[bot] commented on GitHub (Dec 15, 2025):

Here’s what’s happening with Dify 1.11.1:

  • Manual segment indexing: When you add segments, they’re saved with enabled=False and only become searchable after an indexing job runs (like add_document_to_index_task). That’s why you need to close/reopen the knowledge base or trigger re-indexing for retrieval to work. There’s no automatic indexing on insert in this version—this is a known limitation and has been reported by other users as well. Workarounds include re-uploading files or recreating the knowledge base to force re-indexing, but it’s not ideal (source, source, thread).

  • Metadata search and zero values: Metadata filtering uses SQL expressions to convert JSON fields to string or float for searching. If zero values aren’t handled correctly in the filtering logic, searches for 0 may fail. Type consistency (number vs string) depends on both frontend input and backend processing, and bugs with metadata typing and zero-value handling have been reported and partially fixed in newer versions (source). There’s no robust fix in 1.11.1, so you may see inconsistent results.

  • Image retrieval in Simple QA: Image attachments are stored and linked, but reliable retrieval and rendering of image-containing segments in workflows is only fully supported in versions after PR #27793 (December 2025). In 1.11.1, the backend strips markdown image tags and stores files as metadata, but the frontend doesn’t render them, so you won’t see images in recall results. There’s no workaround for this in 1.11.1—upgrading is required for full multimodal/image support (thread, PR, PR).

If you want these features to work as expected, you’ll need to upgrade to a newer version (after PR #27793). When upgrading, watch out for breaking changes in authentication, log retrieval, and knowledge base migration steps. Let me know if you want technical details or upgrade guidance.

To reply, just mention @dosu.

How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other

Chat with Dosu Open in Cursor Join Discord Share on X

@dosubot[bot] commented on GitHub (Dec 15, 2025): <!-- Answer --> Here’s what’s happening with Dify 1.11.1: - **Manual segment indexing:** When you add segments, they’re saved with `enabled=False` and only become searchable after an indexing job runs (like `add_document_to_index_task`). That’s why you need to close/reopen the knowledge base or trigger re-indexing for retrieval to work. There’s no automatic indexing on insert in this version—this is a known limitation and has been reported by other users as well. Workarounds include re-uploading files or recreating the knowledge base to force re-indexing, but it’s not ideal ([source](https://github.com/langgenius/dify/blob/80c74cf725aeb6055b4eb03aabacf82aaec907a3/api/core/rag/docstore/dataset_docstore.py#L13-L254), [source](https://github.com/langgenius/dify/blob/80c74cf725aeb6055b4eb03aabacf82aaec907a3/api/tasks/add_document_to_index_task.py#L21-L130), [thread](https://github.com/langgenius/dify/issues/27798)). - **Metadata search and zero values:** Metadata filtering uses SQL expressions to convert JSON fields to string or float for searching. If zero values aren’t handled correctly in the filtering logic, searches for `0` may fail. Type consistency (number vs string) depends on both frontend input and backend processing, and bugs with metadata typing and zero-value handling have been reported and partially fixed in newer versions ([source](https://github.com/langgenius/dify/blob/80c74cf725aeb6055b4eb03aabacf82aaec907a3/api/core/rag/retrieval/dataset_retrieval.py#L259-L1455)). There’s no robust fix in 1.11.1, so you may see inconsistent results. - **Image retrieval in Simple QA:** Image attachments are stored and linked, but reliable retrieval and rendering of image-containing segments in workflows is only fully supported in versions after PR #27793 (December 2025). In 1.11.1, the backend strips markdown image tags and stores files as metadata, but the frontend doesn’t render them, so you won’t see images in recall results. There’s no workaround for this in 1.11.1—upgrading is required for full multimodal/image support ([thread](https://github.com/langgenius/dify/issues/29485), [PR](https://github.com/langgenius/dify/pull/29115), [PR](https://github.com/langgenius/dify/pull/29431)). If you want these features to work as expected, you’ll need to upgrade to a newer version (after PR #27793). When upgrading, watch out for breaking changes in authentication, log retrieval, and knowledge base migration steps. Let me know if you want technical details or upgrade guidance. <!-- Dosu Comment Footer --> *To reply, just mention [@dosu](https://go.dosu.dev/dosubot).* --- <sup>How did I do? [Good](https://app.dosu.dev/response-feedback/06b93dff-5309-44b0-90d4-beef07d23807?feedback_type=great_response) | [Irrelevant](https://app.dosu.dev/response-feedback/06b93dff-5309-44b0-90d4-beef07d23807?feedback_type=irrelevant_answer) | [Incorrect](https://app.dosu.dev/response-feedback/06b93dff-5309-44b0-90d4-beef07d23807?feedback_type=incorrect_sources) | [Verbose](https://app.dosu.dev/response-feedback/06b93dff-5309-44b0-90d4-beef07d23807?feedback_type=too_verbose) | [Hallucination](https://app.dosu.dev/response-feedback/06b93dff-5309-44b0-90d4-beef07d23807?feedback_type=hallucination) | [Report 🐛](https://app.dosu.dev/response-feedback/06b93dff-5309-44b0-90d4-beef07d23807?feedback_type=bug_report) | [Other](https://app.dosu.dev/response-feedback/06b93dff-5309-44b0-90d4-beef07d23807?feedback_type=other)</sup> [![Chat with Dosu](https://dosu.dev/dosu-chat-badge.svg)](https://app.dosu.dev/6088a4f4-073a-43f3-b3d4-4f06fb0c05a0/ask?utm_source=github)&nbsp;[![Open in Cursor](https://dosu.dev/dosu-in-cursor.svg)](https://cursor.com/link/prompt?text=Here%E2%80%99s%20what%E2%80%99s%20happening%20with%20Dify%201.11.1%3A%0A%0A-%20%2A%2AManual%20segment%20indexing%3A%2A%2A%20When%20you%20add%20segments%2C%20they%E2%80%99re%20saved%20with%20%60enabled%3DFalse%60%20and%20only%20become%20searchable%20after%20an%20indexing%20job%20runs%20%28like%20%60add_document_to_index_task%60%29.%20That%E2%80%99s%20why%20you%20need%20to%20close/reopen%20the%20knowledge%20base%20or%20trigger%20re-indexing%20for%20retrieval%20to%20work.%20There%E2%80%99s%20no%20automatic%20indexing%20on%20insert%20in%20this%20version%E2%80%94this%20is%20a%20known%20limitation%20and%20has%20been%20reported%20by%20other%20users%20as%20well.%20Workarounds%20include%20re-uploading%20files%20or%20recreating%20the%20knowledge%20base%20to%20force%20re-indexing%2C%20but%20it%E2%80%99s%20not%20ideal%20%28%5Bsource%5D%28https%3A//github.com/langgenius/dify/blob/80c74cf725aeb6055b4eb03aabacf82aaec907a3/api/core/rag/docstore/dataset_docstore.py%23L13-L254%29%2C%20%5Bsource%5D%28https%3A//github.com/langgenius/dify/blob/80c74cf725aeb6055b4eb03aabacf82aaec907a3/api/tasks/add_document_to_index_task.py%23L21-L130%29%2C%20%5Bthread%5D%28https%3A//github.com/langgenius/dify/issues/27798%29%29.%0A%0A-%20%2A%2AMetadata%20search%20and%20zero%20values%3A%2A%2A%20Metadata%20filtering%20uses%20SQL%20expressions%20to%20convert%20JSON%20fields%20to%20string%20or%20float%20for%20searching.%20If%20zero%20values%20aren%E2%80%99t%20handled%20correctly%20in%20the%20filtering%20logic%2C%20searches%20for%20%600%60%20may%20fail.%20Type%20consistency%20%28number%20vs%20string%29%20depends%20on%20both%20frontend%20input%20and%20backend%20processing%2C%20and%20bugs%20with%20metadata%20typing%20and%20zero-value%20handling%20have%20been%20reported%20and%20partially%20fixed%20in%20newer%20versions%20%28%5Bsource%5D%28https%3A//github.com/langgenius/dify/blob/80c74cf725aeb6055b4eb03aabacf82aaec907a3/api/core/rag/retrieval/dataset_retrieval.py%23L259-L1455%29%29.%20There%E2%80%99s%20no%20robust%20fix%20in%201.11.1%2C%20so%20you%20may%20see%20inconsistent%20results.%0A%0A-%20%2A%2AImage%20retrieval%20in%20Simple%20QA%3A%2A%2A%20Image%20attachments%20are%20stored%20and%20linked%2C%20but%20reliable%20retrieval%20and%20rendering%20of%20image-containing%20segments%20in%20workflows%20is%20only%20fully%20supported%20in%20versions%20after%20PR%20%2327793%20%28December%202025%29.%20In%201.11.1%2C%20the%20backend%20strips%20markdown%20image%20tags%20and%20stores%20files%20as%20metadata%2C%20but%20the%20frontend%20doesn%E2%80%99t%20render%20them%2C%20so%20you%20won%E2%80%99t%20see%20images%20in%20recall%20results.%20There%E2%80%99s%20no%20workaround%20for%20this%20in%201.11.1%E2%80%94upgrading%20is%20required%20for%20full%20multimodal/image%20support%20%28%5Bthread%5D%28https%3A//github.com/langgenius/dify/issues/29485%29%2C%20%5BPR%5D%28https%3A//github.com/langgenius/dify/pull/29115%29%2C%20%5BPR%5D%28https%3A//github.com/langgenius/dify/pull/29431%29%29.%0A%0AIf%20you%20want%20these%20features%20to%20work%20as%20expected%2C%20you%E2%80%99ll%20need%20to%20upgrade%20to%20a%20newer%20version%20%28after%20PR%20%2327793%29.%20When%20upgrading%2C%20watch%20out%20for%20breaking%20changes%20in%20authentication%2C%20log%20retrieval%2C%20and%20knowledge%20base%20migration%20steps.%20Let%20me%20know%20if%20you%20want%20technical%20details%20or%20upgrade%20guidance.)&nbsp;[![Join Discord](https://img.shields.io/badge/join-5865F2?logo=discord&logoColor=white&label=)](https://go.dosu.dev/discord-bot)&nbsp;[![Share on X](https://img.shields.io/badge/X-share-black)](https://twitter.com/intent/tweet?text=%40dosu_ai%20helped%20me%20solve%20this%20issue!&url=https%3A//github.com/langgenius/dify/issues/29663)
yindo commented 2026-02-21 20:10:26 -05:00
Author
Owner
Copy Link
Copy Source

@froukee commented on GitHub (Dec 15, 2025):

Image retrieval in Simple QA: The problem is that neither text nor images can be retrieved, rather than a front-end rendering issue.

@froukee commented on GitHub (Dec 15, 2025): Image retrieval in Simple QA: The problem is that neither text nor images can be retrieved, rather than a front-end rendering issue.
yindo commented 2026-02-21 20:10:26 -05:00
Author
Owner
Copy Link
Copy Source

@dosubot[bot] commented on GitHub (Dec 15, 2025):

This is a backend limitation in Dify 1.11.1: when you add Simple QA segments with images, neither text nor images are retrievable because segments are only available after successful indexing, and image retrieval requires the dataset to be set as multimodal with high_quality indexing. If indexing fails or the segment is disabled, nothing is returned—even though the data is stored. There’s no workaround for this in 1.11.1; full support for retrieving both text and images in Simple QA was only added after PR #27793 (December 2025). Upgrading to a newer version is required for reliable multimodal retrieval. If you want to troubleshoot further, check the segment/document status in the UI and backend logs (e.g., /app/logs/server.log) for indexing errors or misconfigurations. More details: 1 2.

To reply, just mention @dosu.

How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other

Chat with Dosu Open in Cursor Join Discord Share on X

@dosubot[bot] commented on GitHub (Dec 15, 2025): <!-- Answer --> This is a backend limitation in Dify 1.11.1: when you add Simple QA segments with images, neither text nor images are retrievable because segments are only available after successful indexing, and image retrieval requires the dataset to be set as multimodal with high_quality indexing. If indexing fails or the segment is disabled, nothing is returned—even though the data is stored. There’s no workaround for this in 1.11.1; full support for retrieving both text and images in Simple QA was only added after PR #27793 (December 2025). Upgrading to a newer version is required for reliable multimodal retrieval. If you want to troubleshoot further, check the segment/document status in the UI and backend logs (e.g., `/app/logs/server.log`) for indexing errors or misconfigurations. More details: [1](https://github.com/langgenius/dify/issues/29485) [2](https://github.com/langgenius/dify/blob/80c74cf725aeb6055b4eb03aabacf82aaec907a3/api/core/rag/docstore/dataset_docstore.py#L62-L254). <!-- Dosu Comment Footer --> *To reply, just mention [@dosu](https://go.dosu.dev/dosubot).* --- <sup>How did I do? [Good](https://app.dosu.dev/response-feedback/49dd08a1-a094-4dd7-9787-0921181b4a10?feedback_type=great_response) | [Irrelevant](https://app.dosu.dev/response-feedback/49dd08a1-a094-4dd7-9787-0921181b4a10?feedback_type=irrelevant_answer) | [Incorrect](https://app.dosu.dev/response-feedback/49dd08a1-a094-4dd7-9787-0921181b4a10?feedback_type=incorrect_sources) | [Verbose](https://app.dosu.dev/response-feedback/49dd08a1-a094-4dd7-9787-0921181b4a10?feedback_type=too_verbose) | [Hallucination](https://app.dosu.dev/response-feedback/49dd08a1-a094-4dd7-9787-0921181b4a10?feedback_type=hallucination) | [Report 🐛](https://app.dosu.dev/response-feedback/49dd08a1-a094-4dd7-9787-0921181b4a10?feedback_type=bug_report) | [Other](https://app.dosu.dev/response-feedback/49dd08a1-a094-4dd7-9787-0921181b4a10?feedback_type=other)</sup> [![Chat with Dosu](https://dosu.dev/dosu-chat-badge.svg)](https://app.dosu.dev/6088a4f4-073a-43f3-b3d4-4f06fb0c05a0/ask?utm_source=github)&nbsp;[![Open in Cursor](https://dosu.dev/dosu-in-cursor.svg)](https://cursor.com/link/prompt?text=This%20is%20a%20backend%20limitation%20in%20Dify%201.11.1%3A%20when%20you%20add%20Simple%20QA%20segments%20with%20images%2C%20neither%20text%20nor%20images%20are%20retrievable%20because%20segments%20are%20only%20available%20after%20successful%20indexing%2C%20and%20image%20retrieval%20requires%20the%20dataset%20to%20be%20set%20as%20multimodal%20with%20high_quality%20indexing.%20If%20indexing%20fails%20or%20the%20segment%20is%20disabled%2C%20nothing%20is%20returned%E2%80%94even%20though%20the%20data%20is%20stored.%20There%E2%80%99s%20no%20workaround%20for%20this%20in%201.11.1%3B%20full%20support%20for%20retrieving%20both%20text%20and%20images%20in%20Simple%20QA%20was%20only%20added%20after%20PR%20%2327793%20%28December%202025%29.%20Upgrading%20to%20a%20newer%20version%20is%20required%20for%20reliable%20multimodal%20retrieval.%20If%20you%20want%20to%20troubleshoot%20further%2C%20check%20the%20segment/document%20status%20in%20the%20UI%20and%20backend%20logs%20%28e.g.%2C%20%60/app/logs/server.log%60%29%20for%20indexing%20errors%20or%20misconfigurations.%20More%20details%3A%20%5B1%5D%28https%3A//github.com/langgenius/dify/issues/29485%29%20%5B2%5D%28https%3A//github.com/langgenius/dify/blob/80c74cf725aeb6055b4eb03aabacf82aaec907a3/api/core/rag/docstore/dataset_docstore.py%23L62-L254%29.)&nbsp;[![Join Discord](https://img.shields.io/badge/join-5865F2?logo=discord&logoColor=white&label=)](https://go.dosu.dev/discord-bot)&nbsp;[![Share on X](https://img.shields.io/badge/X-share-black)](https://twitter.com/intent/tweet?text=%40dosu_ai%20helped%20me%20solve%20this%20issue!&url=https%3A//github.com/langgenius/dify/issues/29663)
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
Knowledge Pipeline
Proposal
WIP
cloud
dependencies
feat:webapp
good first issue
outdated
performance
pull-request
Mirrored from GitHub Pull Request
 python:uv
refactor
size:L
stale
status: accepting prs
status: blocked
status: in discussion
stt & tts
vulnerablity
web
wont-fix
🌊 feat:workflow
🌚 invalid
🐞 bug
👻 feat:rag
💪 enhancement
📚 documentation
🔨 feat:tools
🙋‍♂️ question
🙏 help wanted
🤔 cant-reproduce
🤖 feat:agent
No Label 🐞 bug
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
yindo
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: langgenius/dify#21048
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?