Why are the parallel branches not executed in parallel？

[Vurtnel]

Self Checks

• I have searched for existing issues search for existing issues, including closed ones.
• I confirm that I am using English to submit this report (我已阅读并同意 Language Policy).
• [FOR CHINESE USERS] 请务必使用英文提交 Issue，否则会被关闭。谢谢！:)
• Please do not modify this template :) and fill in all the required fields.

Content

workflow:

exexute process:

[dosubot[bot]]

By default, Dify executes workflow nodes sequentially—even if your diagram shows parallel branches—because the worker pool is set to only one worker. To enable true parallel execution, set GRAPH_ENGINE_MIN_WORKERS and GRAPH_ENGINE_MAX_WORKERS in your .env file (for example, GRAPH_ENGINE_MIN_WORKERS=4 and GRAPH_ENGINE_MAX_WORKERS=10), then restart your Docker containers. This allows multiple branches to run at the same time, up to the worker pool limit. Parallelism is controlled only by these environment variables, not by workflow YAML or UI settings.

If you still see sequential execution after updating the settings, check that the environment variables are actually set inside your running containers, that Docker has enough CPU/memory allocated, and that your workflow design doesn’t have hidden dependencies or nodes with multiple incoming edges (which Dify’s engine doesn’t fully support for parallelism). Also, technical bottlenecks like synchronous APIs or blocking I/O can prevent true parallelism, and sometimes the UI may show sequential execution even if parallelism is happening under the hood. For more details and troubleshooting steps, see this discussion.

To reply, just mention @dosu.

---

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

   

[Vurtnel]

i have set the GRAPH_ENGINE_MAX_WORKERS=10, but it still not executed in parallel

[dosubot[bot]]

Setting only GRAPH_ENGINE_MAX_WORKERS is not enough for parallel execution in Dify. You also need to set GRAPH_ENGINE_MIN_WORKERS to a value greater than 1 (for example, 2 or 4) in your .env file—this controls how many workers are actually started for small workflows. With the default min_workers=1, even a high max_workers value won't help unless the workflow is large or the queue backs up, which isn't the case for two parallel branches.

Try adding:

GRAPH_ENGINE_MIN_WORKERS=2
GRAPH_ENGINE_MAX_WORKERS=10

and restart your Docker containers. For best results, also set GRAPH_ENGINE_SCALE_UP_THRESHOLD=1 to ensure scaling happens even with a small number of parallel nodes.

If you still see sequential execution, double-check that the environment variables are visible inside your running container (e.g., with docker exec -it <container> env | grep GRAPH_ENGINE), that your workflow doesn't have hidden dependencies between branches, and that Docker has enough CPU/memory allocated for multiple workers. More details and troubleshooting steps are available in this discussion and here.

To reply, just mention @dosu.

---

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