About docker-compose deployment issues

[weikecloud]

Self Checks

• This is only for bug report, if you would like to ask a quesion, 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 (我已阅读并同意 Language Policy).
• Pleas do not modify this template :) and fill in all the required fields.

Dify version

0.6.3

Cloud or Self Hosted

Self Hosted (Docker)

Steps to reproduce

I use containers to deploy each service separately. After entering the web and having a conversation, it will cause the api service to error and restart.

✔️ Expected Behavior

Talk to the model normally

❌ Actual Behavior

/entrypoint.sh: line 27: 49 Killed gunicorn --bind "${DIFY_BIND_ADDRESS:-0.0.0.0}:${DIFY_PORT:-5001}" --workers ${SERVER_WORKER_AMOUNT:-1} --worker-class ${SERVER_WORKER_CLASS:-gevent} --timeout ${GUNICORN_TIMEOUT:-200} --preload app:app

[dosubot[bot]]

The error you're encountering with the gunicorn server being killed can be attributed to several potential causes. Here's a focused approach to troubleshoot and possibly resolve the issue:

1. Check for Insufficient Memory: Ensure your deployment environment has enough memory, especially if you're running multiple workers. The system might kill processes to free up memory if it's running low.

2. Review gunicorn Configuration: Adjust the gunicorn settings in your entrypoint.sh script. Consider reducing the number of workers (SERVER_WORKER_AMOUNT) or increasing the timeout (GUNICORN_TIMEOUT, recommended to set at 360 for longer SSE connection time) to see if it impacts stability.

3. Examine System Logs: Look for any out-of-memory errors or signals (SIGKILL, SIGTERM) that might indicate why the process is being terminated.

4. Worker Class Compatibility: Since you're using gevent as the worker class, ensure your application and its dependencies are compatible with asynchronous workers. If not, consider switching to a different worker class.

5. Application Preloading: The --preload option can cause issues if your application isn't designed for it. If you suspect this, try disabling it to see if the situation improves.

6. Deployment Environment Details: Providing more information about your operating system, hardware specifications, and any other relevant deployment details can help in diagnosing the issue more accurately.

By focusing on these areas, you should be able to narrow down the cause of the gunicorn process termination and find a suitable solution.

Details

Sources

• api/docker/entrypoint.sh
• docker/docker-compose.yaml
• docker/docker-compose.middleware.yaml
• zh_CN/getting-started/install-self-hosted/environments.md
• en/getting-started/install-self-hosted/docker-compose.md
• en/getting-started/install-self-hosted/environments.md

About Dosu
This response is meant to be useful and save you time. It is not meant to be a precise solution, but rather a starting point for your own research.

Help me be more useful! Please leave a 👍 if this is helpful and 👎 if it is irrelevant.

To continue the conversation, mention @dosu.