Chatflow's concurrent nodes cannot run tasks in parallel.

[RajanZhan]

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

1.9

Cloud or Self Hosted

Self Hosted (Docker)

Steps to reproduce

I built a chatflow with concurrent nodes, as shown in Figure 1, 2, and 3, and ran and debugged it directly. I expect nodes 1, 2, and 3 to execute concurrently, but the actual execution process is sequential. How can I configure it to achieve parallel execution?

✔️ Expected Behavior

I expect nodes 1, 2, and 3 to execute concurrently。

❌ Actual Behavior

the actual execution process is sequential.

chatflow.yml

[dosubot]

Hi @RajanZhan! I'm Dosu and I’m helping the dify team.

Dify 1.9 does support parallel execution of concurrent nodes, but by default it runs all nodes sequentially because the worker pool is set to only one worker. To enable true parallel execution in your chatflow, set the following environment variables in your .env file (in your Dify deployment root):

GRAPH_ENGINE_MIN_WORKERS=4
GRAPH_ENGINE_MAX_WORKERS=10

Then restart your Docker containers for the changes to take effect. This will allow multiple nodes to run at the same time, up to the thread pool limit (default max is 10) details.

Parallelism is not controlled in chatflow.yml—it’s only managed by these environment variables. If you still see sequential execution after updating the settings, check your Docker resource limits (CPU/memory) and make sure your workflow design doesn’t have unnecessary dependencies between nodes details.

Note: If your workflow has nodes with multiple incoming edges from parallel branches, Dify’s engine may not execute them as expected due to design limitations—node reuse in parallel branches is not fully supported details.

For more info, see the relevant issues: issue 26947, issue 29112, discussion 26166.

To reply, just mention @dosu.

---

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

   

[RajanZhan]

@dos

Hi @RajanZhan! I'm Dosu and I’m helping the dify team.

Dify 1.9 does support parallel execution of concurrent nodes, but by default it runs all nodes sequentially because the worker pool is set to only one worker. To enable true parallel execution in your chatflow, set the following environment variables in your .env file (in your Dify deployment root):

GRAPH_ENGINE_MIN_WORKERS=4
GRAPH_ENGINE_MAX_WORKERS=10

Then restart your Docker containers for the changes to take effect. This will allow multiple nodes to run at the same time, up to the thread pool limit (default max is 10) details.

Parallelism is not controlled in chatflow.yml—it’s only managed by these environment variables. If you still see sequential execution after updating the settings, check your Docker resource limits (CPU/memory) and make sure your workflow design doesn’t have unnecessary dependencies between nodes details.

Note: If your workflow has nodes with multiple incoming edges from parallel branches, Dify’s engine may not execute them as expected due to design limitations—node reuse in parallel branches is not fully supported details.

For more info, see the relevant issues: issue 26947, issue 29112, discussion 26166.

To reply, just mention @dosu.

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

   

你好@RajanZhan我是Dosu，我正在协助 dify 团队。

Dify 1.9 支持并发节点的并行执行，但默认情况下所有节点按顺序运行，因为工作池仅设置为一个工作节点。要在聊天流程中启用真正的并行执行，请在配置.env文件（位于 Dify 部署根目录）中设置以下环境变量：

GRAPH_ENGINE_MIN_WORKERS=4
GRAPH_ENGINE_MAX_WORKERS=10

然后重启 Docker 容器以使更改生效。这将允许多个节点同时运行，最多可达线程池限制（默认最大值为 10） 详情。

并行性并非由 Docker 控制，而是由这些环境变量管理。如果在更新设置后仍然看到顺序执行，请检查 Docker 资源限制（CPU/内存），并确保工作流设计中节点chatflow.yml之间没有不必要的依赖关系 。

注意：如果您的工作流中存在来自并行分支的多个传入边的节点，由于设计限制，Dify 的引擎可能无法按预期执行它们——并行分支中的节点重用尚未完全支持 （详情）。

有关更多信息，请参阅相关问题：问题 26947、问题 29112、讨论 26166。

回复时只需提及@dosu即可。

我做得怎么样？好|无关|错误|冗长|幻觉|举报 🐛 |其他

   

I have already configured and restarted Docker, but it's still running sequentially. My environment is a Windows 11 Docker environment. Please advise on the cause.

[dosubot]

Here are the most likely reasons why parallel execution still isn't working in your Windows 11 Docker setup:

1. Environment variables not applied: Double-check that GRAPH_ENGINE_MIN_WORKERS and GRAPH_ENGINE_MAX_WORKERS are actually set inside your running containers. You can verify this by running:

   docker exec -it <your_api_or_worker_container_name> env | grep GRAPH_ENGINE
   

   If they're missing, make sure your .env file is in the correct directory and that you restarted the containers after editing it. The Dify codebase loads these variables directly from the container environment—there's no platform-specific logic for Windows, so they should work the same as on Linux cite.

2. Docker Desktop resource limits: On Windows, Docker Desktop limits CPU and memory via its settings panel. If you have only 1–2 CPUs or low memory allocated, parallelism will be restricted even if the worker pool is configured for concurrency cite.

3. Workflow structure: If your chatflow has hidden dependencies or nodes with multiple incoming edges from parallel branches, Dify may still execute them sequentially due to engine limitations cite.

4. Scaling logic: The worker pool only increases workers if the ready queue depth exceeds GRAPH_ENGINE_SCALE_UP_THRESHOLD. If your workflow enqueues nodes one after another, extra workers may never be triggered cite.

5. Technical bottlenecks: Synchronous model APIs, blocking I/O, or Flask context issues can also prevent true parallelism cite.

6. UI display: Sometimes, parallel execution is happening but the UI shows it as sequential. Tools like Langfuse's waterfall view can help confirm real execution order cite.

If you're running Docker on Windows filesystems, try copying your project into WSL and running Docker from there to avoid permission or environment propagation issues cite.

If none of these resolve it, consider upgrading to a newer Dify version, as several parallel execution bugs were fixed after 1.9.0 cite.

To reply, just mention @dosu.

---

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