Nodes stay in "Running" state when using "Run this step"

[MitraSafarinejad]

Self Checks

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

Dify version

0.14.1

Cloud or Self Hosted

Self Hosted (Docker)

Steps to reproduce

When testing individual nodes using the "Run this step" feature, the node remains in the "Running" state indefinitely and does not complete. This issue occurs with nodes like Question Classifier, Knowledge Retrieval, and others. However, when running the entire flow in a chatflow, the nodes execute correctly without any issues.

✔️ Expected Behavior

The node should complete its execution and display the result or an error message within a reasonable amount of time.

❌ Actual Behavior

The node remains in the "Running" state indefinitely without completing or providing feedback. However, the node works correctly when the entire chatflow is executed.

[dosubot[bot]]

I found a similar open issue regarding nodes staying in the "RUNNING" state. The suggested solutions include:

1. Event Handling: Ensure that event handlers for node execution, such as onNodeStarted and onNodeFinished, are correctly implemented. If these handlers are not triggered, it might cause the node to remain in a "RUNNING" state.

2. Parallel Node Execution: Check for synchronization issues or deadlocks, especially if your setup involves parallel nodes. Ensure nodes are configured correctly for parallel execution and that there are no dependencies causing a deadlock.

3. API 'event:ping' Events: Continuous "event:ping" responses might indicate the workflow is stuck in a loop or waiting for a condition that is never met. Ensure all conditions and loops have proper exit criteria.

4. Logging and Debugging: Use Dify's logging features to gain insights into node executions. Even if no errors are visible, logs might provide clues about where the process is getting stuck.

5. Version-Specific Issues: Since you're using version 0.14.1, check for any known issues or updates related to this version that might affect node execution ^[1].

These steps might help identify the root cause of the issue with the "Run this step" feature.

To continue talking to Dosu, mention @dosu.

---

^{Help Dosu learn! Give it feedback: Great Response | Irrelevant Answer | Incorrect Sources | Too Verbose | Hallucination | Other}

[MitraSafarinejad]

I have discovered that this issue causes an even more critical problem. When I use the "Run this step" feature, and the node stays in the "Running" state, it continuously calls APIs, leading to infinite API requests. For example, when testing an LLM Node, it keeps sending requests to the OpenAI API without stopping, resulting in significant financial costs. This makes the issue highly urgent and requires immediate attention. Please prioritize this as it is causing substantial damage. @crazywoola

[crazywoola]

@MitraSafarinejad We will take a look at this ASAP. Can you attach a screenshot or the DSL of the workflow? I am not able to see the configuration of the node here. I can not reproduce this problem in our cloud env.

[MitraSafarinejad]

@crazywoola Thank you for your attention to this issue.

Here are the nodes and the chatflow I am using. It’s a simple flow, so I hope this helps you understand the configuration and reproduce the problem. Additionally, I want to note that I am running the system using Docker, not the cloud environment.

Please let me know if you need any additional details or further clarification.

Looking forward to your updates!

[laipz8200]

Hi @MitraSafarinejad! I tried to create a similar chatflow with you in version 0.14.1, using the same model, but I can't reproduce this issue. Could you provide more information? Maybe you could copy the API logs from your Docker container when this issue occurs.

[MitraSafarinejad]

Hi @laipz8200,

Thanks for following up! Below, I’ve included the logs generated when I ran the LLM nodes, along with the DSL file exported from my chatflow for your reference.

log.txt dsl.txt

[MitraSafarinejad]

hello, I updated my dify version to 0.14.2 and the problem does not exist anymore

[laipz8200]

Hi @MitraSafarinejad. Thank you for your feedback!

[HanXM715]

@laipz8200 @crazywoola My current version is 0.14.2 and I have a similar problem, also calling the API constantly, resulting in infinite API requests

[HanXM715]

[tkysi-mi]

I am encountering a similar issue in Cloud version right now...