Langgraph process uses always 100% CPU

[tisDDM]

When I start up the Dev-Server with langgraph up or start the docker manually the server process always speeds up to 100% usage (one core).

Has anyone encountered similiar? Any known solutions?

[marc-hughes]

I also experience this while running langgraph up --watch --port 54367 inside an ubuntu based WSL on Windows 11.

[TomTom101]

Same, it then crashes after a while

[bracesproul]

Hi @tisDDM @marc-hughes @TomTom101 we recently released an in memory server for LangGraph.js which is now the recommended method for running LangGraph.js locally for testing/development.

You can use the in mem server by running the yarn dev:server script!

[tisDDM]

Hi @bracesproul ,

Thank you for your work in this project and your comment to the issue.

I stopped working with your canvas - which is a very great idea indeed. But there are reasons why I do not put any effort into using it.

1. The canvas itself is open source and under an open license. But langgraph isn't. Hard to run it w/o legal issues. And why should I just take part in developing without using?
2. Supabase is annoying. Why is the authentication (and more) part of the project, especially in this way? Normally this shall be handled in an other instance of the infrastructure. Furthermore they sent annoying advertisements. It stopped when the hibernated my account due to 6 days(!!!!) of inactivity. Great.
3. Code quality could be a lot improved. Found tons of hardcoded stuff, that needed fixing for use in a multi-container environment. Not talking about the strange hassle when implementing new LLMs
4. Even langgraph runs unstable at least locally. Its hard and unintuitive to work with this product.

I am a very experienced developer and I spent a few month in the langchain universe.

I already have quit and be heading in direction of CrewAI.