v0.14.0

[laipz8200]

✨ Discover What's New in v0.14.0 ✨

We’re excited to bring you the latest v0.14.0 update. Let's dive straight into some of the standout features and improvements. As always, we extend a special thanks to our contributors!

⚡ New Features

🔗 Improved Error Handling

Enhanced functionality for workflows and chatflows allows nodes like LLM and HTTP requests to continue operations even when errors occur. Big shoutout to @zxhlyh and @Nov1c444 for their work in #11474 and #11458.

🎨 Revamped User Interface

Creation	Exploration

With the fresh entry point for app creation and template exploration, user experience becomes more intuitive, assisting in quick and efficient startup. Thanks to @douxc for this in #10847.

📁 File Previews Directly in Chat

Now, video, audio, and PDF file previews are accessible right within the chat, simplifying on-the-go information access. Credit goes to @charli117 in #11466.

Learn more in our blog

🧠 Model Updates

🌍 Gemini 2.0 and Flash Exp in Google

This enhancement by @hjlarry and @wangiii, found in #11570 and #11604, boosts overall performance.

🔢 Grok-2-1212 and Grok-2-Vision-1212 in XAI

Optimizing the model functionality, thanks to @hjlarry in #11672.

🧙 QwQ and Llama3.3 Models in Siliconflow

Expands compatibility and versatility, contributed by @orangeclk in #11492.

📷 Abab6.5t-Chat Model in Minimax

Broaden available models by @acelyc111 in #11484.

🔧 Other Highlights

🏦 Storage Integration

Features integration of Apache OpenDAL storage for enhanced system-wide storage capabilities. Work done by @laipz8200 in #11508.

📖 Knowledge Base Enhancements

Added support for MDX files in the Knowledge Base to improve content flexibility. Credit to @yihong0618 in #11565.

🔧 Logging and Dev Tools

Logs now include trace IDs for smoother debugging, thanks to @luckylhb90 in #11599. Plus, @fujita-h added handy Docker commands in #11616.

🔧 Bug Fixes

Various minor bugs have been addressed, including iteration node errors and proper release of rate-restricting requests. Thanks to @yihong0618, @liuzhenghua, and others for resolving these issues in #11504, #11540, and more.

🌈 Additional Enhancements

Lindorm VDB support and configurable max submit count, thanks to @AlwaysBluer and @kazuhisa-wada in #11574 and #11673.

Every update in this release reinforces the platform's robustness and adaptability—many thanks to all contributors! Let's continue building this powerful community together! 🚀

---

Upgrade Guide

Docker compose deployments

Warning

The files in the docker-legacy directory will soon stop being maintained and will be removed from the repository. If you are still using them, please switch to the new version as soon as possible.

1. Back up your customized docker-compose YAML file (optional)

   cd docker
   cp docker-compose.yaml docker-compose.yaml.$(date +%s).bak

2. Get the latest code from the main branch

   git checkout main
   git pull origin main

3. Stop the service，Command, please execute in the docker directory

   docker compose down

4. Back up data

   tar -cvf volumes-$(date +%s).tgz volumes

5. Upgrade services

   docker compose up -d

Source Code deployments

1. Stop API server, Worker and Web frontend Server.

2. Get the latest code from the release branch:

   git checkout 0.14.0

3. Update Python dependencies:

   cd api
   poetry install

4. Then, let's run the migration script:

   poetry run flask db upgrade

5. Finally, run API server, Worker and Web frontend Server again.

---

What's Changed

• fix: add the missing abab6.5t-chat model of Minimax by @acelyc111 in fix: add the missing abab6.5t-chat model of Minimax #11484
• feat:add apollo configuration to load env file by @huanshare in feat:add apollo configuration to load env file #11210
• feat: add siliconflow qwq and llama3.3 model by @orangeclk in feat: add siliconflow qwq and llama3.3 model #11492
• fix(app_generator_service): overload type hints by @laipz8200 in fix(app_generator_service): overload type hints #11507
• chore : fix translation Typo in ko-KR localization by @JungHyunMoon in chore : fix translation Typo in ko-KR localization #11509
• chore: fix typo in Japanese localization by @barabicu in chore: fix typo in Japanese localization #11502
• Msg file preview by @charli117 in Msg file preview #11466
• chore: LOCAL_FILE also try to use remote_url as Prompt message by @hjlarry in chore: LOCAL_FILE also try to use remote_url as Prompt message #11443
• Remove the processing of single quote when testing API tools. by @weimzh in Remove the processing of single quote when testing API tools. #11390
• fix: issue 11247 that Completion mode content maybe list or str by @yihong0618 in fix: issue 11247 that Completion mode content maybe list or str #11504
• fix: can not start local by REMOTE_SETTINGS_SOURCE_NAME change it to … by @yihong0618 in fix: can not start local by REMOTE_SETTINGS_SOURCE_NAME change it to … #11535
• Revert "fix: total tokens is wrong which is zero in inter way, close … by @yihong0618 in Revert "fix: total tokens is wrong which is zero in inter way, close … #11536
• fix: better error message for url add external knowledge by @yihong0618 in fix: better error message for url add external knowledge #11537
• feat: Add llama-3.3 models for Groq by @pvoo in feat: Add llama-3.3 models for Groq #11533
• fix: Remove duplicate 'response_format' parameter from model YAML files by @zkytech in fix: Remove duplicate 'response_format' parameter from model YAML files #11531
• [Pixtral] Add new model ; add vision by @Asterovim in [Pixtral] Add new model ; add vision #11231
• chore: update thai lang in app page by @YIXIAO0 in chore: update thai lang in app page #11541
• feat: workflow continue on error by @zxhlyh in feat: workflow continue on error #11474
• Feat: continue on error by @Nov1c444 in Feat: continue on error #11458
• fix: iteration node in parallel mode token count error by @Nov1c444 in fix: iteration node in parallel mode token count error #11539
• feat: integrate opendal storage by @laipz8200 in feat: integrate opendal storage #11508
• fix: workflow continue on error doc link by @zxhlyh in fix: workflow continue on error doc link #11554
• chore: translate i18n files by @github-actions in chore: translate i18n files #11545
• Fix: RateLimit requests were not released when a streaming generation exception occurred by @liuzhenghua in Fix: RateLimit requests were not released when a streaming generation exception occurred #11540
• fix: better opendal tests by @yihong0618 in fix: better opendal tests #11569
• feat: add gemini-2.0-flash-exp by @hjlarry in feat: add gemini-2.0-flash-exp #11570
• Lindorm vdb by @AlwaysBluer in Lindorm vdb #11574
• Feat: dark mode for logs and annotations by @JzoNgKVO in Feat: dark mode for logs and annotations #11575
• fix: change workflow trace id by @ZhouhaoJiang in fix: change workflow trace id #11585
• fix: support mdx files close The knowledge base should support importing mdx files #11557 by @yihong0618 in fix: support mdx files close #11557 #11565
• chore: translate i18n files by @github-actions in chore: translate i18n files #11577
• [ref] use one method to get boto client for aws bedrock by @warren830 in [ref] use one method to get boto client for aws bedrock #11506
• Added new models and Removed the deleted ones for Groq Support newer Groq Models and remove the deprecated ones #11455 by @alokas in Added new models and Removed the deleted ones for Groq #11455 #11456
• fix: name of llama-3.3-70b-specdec by @crazywoola in fix: name of llama-3.3-70b-specdec #11596
• feat(model): add vertex_ai Gemini 2.0 Flash Exp by @wangiii in feat(model): add vertex_ai Gemini 2.0 Flash Exp #11604
• ci: better print version for ruff to check the change by @yihong0618 in ci: better print version for ruff to check the change #11587
• feat(devcontainer): add alias to stop Docker containers by @fujita-h in feat(devcontainer): add alias to stop Docker containers #11616
• Fix/pdf preview in build by @zxhlyh in Fix/pdf preview in build #11621
• chore(dependency): bump gunicorn to 23.0 by @laipz8200 in chore(dependency): bump gunicorn to 23.0 #11560
• fix: split dir for opendal tests by @yihong0618 in fix: split dir for opendal tests #11627
• fix: app log filter value error by @Kevin9703 in fix: app log filter value error #11624
• Feat: new entry point for app creation by @douxc in Feat: new entry point for app creation #10847
• fix: filter bug for keywork cause code can not reach by @yihong0618 in fix: filter bug for keywork cause code can not reach #11666
• fix: remove unnecessary curly braces in wf api doc by @RockChinQ in fix: remove unnecessary curly braces in wf api doc #11658
• fix: rename README filename: Japanese language code is 'JA' by @iwaim in fix: rename README filename: Japanese language code is 'JA' #11651
• fix: langfuse do not have created_at args and fix the typing in the file by @yihong0618 in fix: langfuse do not have created_at args and fix the typing in the file #11648
• Update translate to Japanese: natural Japanese expression by @iwaim in Update translate to Japanese: natural Japanese expression #11647
• fix: better error message for stream by @yihong0618 in fix: better error message for stream #11635
• test: run vdb tests on TiDB Vector with docker in CI tests by @bowenliang123 in test: run vdb tests on TiDB Vector with docker in CI tests #11645
• fix: account.id should account_id by @yihong0618 in fix: account.id should account_id #11628
• fix: ExternalDatasetService.process_external_api wrong args by @yihong0618 in fix: ExternalDatasetService.process_external_api wrong args #11586
• fix: _handle_workflow_run_partial_success args is wrong by @yihong0618 in fix: _handle_workflow_run_partial_success args is wrong #11562
• feat: log add trace id by @luckylhb90 in feat: log add trace id #11599
• chore: translate i18n files by @github-actions in chore: translate i18n files #11639
• fix: remove the unused QueueWorkflowPartialSuccessEvent handle in workflow by @Nov1c444 in fix: remove the unused QueueWorkflowPartialSuccessEvent handle in workflow #11669
• feat: add grok-2-1212 and grok-2-vision-1212 by @hjlarry in feat: add grok-2-1212 and grok-2-vision-1212 #11672
• fix: change http node params from dict to list tuple by @hjlarry in fix: change http node params from dict to list tuple #11665
• fix: deepseek reports an error when using Response Format deepseek reports an error when using Response Format #11677 by @zhaobingshuang in fix: deepseek reports an error when using Response Format #11677 #11678
• Make max_submit_count configurable via Config by @kazuhisa-wada in Make max_submit_count configurable via Config #11673
• fix: volcengine_maas and baichuan message error by @zhongliliu-butterfly in fix: volcengine_maas and baichuan message error #11625
• fix: enhance workflow.tool_published performance by @wangbin77 in fix: enhance workflow.tool_published performance #11640
• fix: tool constant params change cause page crashed by @iamjoel in fix: tool constant params change cause page crashed #11682
• fix: iteration output array type causes always outputting string array by @iamjoel in fix: iteration output array type causes always outputting string array #11686
• fix: workflow continue on error edge color by @zxhlyh in fix: workflow continue on error edge color #11689
• chore: bump version to 0.14.0 by @laipz8200 in chore: bump version to 0.14.0 #11679

New Contributors

• @acelyc111 made their first contribution in fix: add the missing abab6.5t-chat model of Minimax #11484
• @huanshare made their first contribution in feat:add apollo configuration to load env file #11210
• @JungHyunMoon made their first contribution in chore : fix translation Typo in ko-KR localization #11509
• @barabicu made their first contribution in chore: fix typo in Japanese localization #11502
• @weimzh made their first contribution in Remove the processing of single quote when testing API tools. #11390
• @pvoo made their first contribution in feat: Add llama-3.3 models for Groq #11533
• @zkytech made their first contribution in fix: Remove duplicate 'response_format' parameter from model YAML files #11531
• @Asterovim made their first contribution in [Pixtral] Add new model ; add vision #11231
• @alokas made their first contribution in Added new models and Removed the deleted ones for Groq #11455 #11456
• @iwaim made their first contribution in fix: rename README filename: Japanese language code is 'JA' #11651
• @zhongliliu-butterfly made their first contribution in fix: volcengine_maas and baichuan message error #11625
• @wangbin77 made their first contribution in fix: enhance workflow.tool_published performance #11640

Full Changelog: 0.13.2...0.14.0

---

This discussion was created from the release v0.14.0.

[breakstring]

OOM Exception after upgraded to 0.14.0

After upgrading from version 0.13.2 to 0.14.0 using Docker-Compose, the service crashed. I was even unable to SSH into my remote server (Azure VM). After several reboots, I still couldn't connect, but later saw the following message from the serial console of the VM in Azure:

dify login: [ 27.033521] cloud-init[2474]: Cloud-init v. 24.3.1-0ubuntu024.04.2 running 'modules:final' at Mon, 16 Dec 2024 09:24:29 +0000. Up 26.99 seconds.
[ 27.059871] cloud-init[2474]: Cloud-init v. 24.3.1-0ubuntu024.04.2 finished at Mon, 16 Dec 2024 09:24:29 +0000. Datasource DataSourceAzure [seed=/var/lib/waagent]. Up 27.05 seconds
[ 134.663354] Out of memory: Killed process 2249 (celery) total-vm:35212808kB, anon-rss:736240kB, file-rss:0kB, shmem-rss:0kB, UID:0 pgtables:2220kB oom_score_adj:0

dify login: [ 2127.518524] Out of memory: Killed process 21467 (gunicorn) total-vm:35131260kB, anon-rss:727544kB, file-rss:0kB, shmem-rss:0kB, UID:0 pgtables:2200kB oom_score_adj:0

It seems that the programs related to these OOM error messages are all associated with Dify.
I tried upgrading the server configuration from 4C8G to 4C16G, and now I can finally log in to the server normally. From the following commands, we can also see that the API service is now taking up quite a bit of memory. It seems that running on 4C8G would be quite challenging.

[breakstring]

If you modified SERVER_WORKER_AMOUNT, try to keep it set to 1. Normally, docker-api will use 900 - 1500MB of memory.

Oops, I guess I might have changed this parameter.
I remember the .env.example had a description suggesting adjusting it according to the number of CPU cores. However, I forgot what I changed it to.
It seems necessary to remind that it should also be set according to the memory size in the example file.

Thank you so much.

[laipz8200]

Thank you for your reminder, I'll update the suggestion. For Gevent, it should always be 1.

[bowenliang123]

Shall we pin SERVER_WORKER_AMOUNT to 1 for API service?

[laipz8200]

@bowenliang123 We also allow users to use workers other than Gevent. This can be suggested as an option and does not need to be pinned.

[bowenliang123]

Sure, sgtm.