Debug with LEAN CLI - connect ECONNREFUSED 127.0.0.1:5678

Based on the information provided in the extracts, it seems like you're trying to debug your algorithm using the Lean CLI and encountering an issue with the connection to Docker.

The error message "connect ECONNREFUSED 127.0.0.1:5678" suggests that the debugger is unable to connect to the specified port on your localhost. This could be due to a variety of reasons, such as Docker not running, the specified port being blocked or used by another process, or a misconfiguration in your launch.json file.

Here are a few steps you can take to troubleshoot this issue:

1. Ensure Docker is running: Docker needs to be running for the Lean CLI to work. You can check if Docker is running by using the command docker info in your terminal. If Docker is not running, you may need to start it manually.

2. Check if the port is blocked or used by another process: You can use the command lsof -i :5678 in your terminal to check if the port 5678 is being used by another process. If it is, you may need to stop that process or use a different port.

3. Review your launch.json configuration: The launch.json file is used to configure the debugger in Visual Studio Code. Based on the extracts, your configuration seems to be correct, but you may want to double-check it against the documentation.

If you continue to experience issues, you may want to refer to the QuantConnect documentation or ask for help in the QuantConnect forum.

I ran

docker info

and the return showed a bunch of information, indicating Docker is running.

I then ran

lsof -i :5678

and it returned nothing, which I assume means there is nothing currently using of that port.

Not sure what to do.

Based on the information provided in the extracts, it seems like your Docker is running and the port 5678 is not being used by another process, which are both good signs.

One thing to note when running algorithms in a Docker container is that localhost does not point to your computer's localhost. Instead, you should substitute localhost with host.docker.internal if your algorithm needs to connect to other services running on your computer. In other words, instead of connecting to http://localhost:<port>/, connect to http://host.docker.internal:<port>/. This information is mentioned in the Deployment section of the documentation.

However, this might not be the issue as the error message you're seeing typically indicates a problem with the debugger connecting to the Docker container, not the algorithm connecting to other services.

If you're still unable to solve your issues, you are welcome to contact the QuantConnect engineers by opening an issue in the QuantConnect/lean-cli repository on GitHub. Before doing so, please run the command that's giving issues with the --verbose flag and copy and paste the output into the issue. The --verbose flag enables debug messages to be printed, which makes it easier for the engineers to help you. This information is mentioned in the Troubleshooting section of the documentation.

After replacing the

"host": "localhost",

line in the launch.json file with

"host": "host.docker.internal",

instead of:

connect ECONNREFUSED 127.0.0.1:5678

I received the error message:

getaddrinfo ENOTFOUND host.docker.internal

Not sure what to do.

The error message getaddrinfo ENOTFOUND host.docker.internal suggests that the system is unable to resolve the hostname host.docker.internal. This hostname is typically used to connect to services running on the host from within a Docker container.

However, it's important to note that host.docker.internal is currently only guaranteed to work on Windows and Mac systems. If you're running Docker on a Linux system, this hostname may not be recognized, which could be the cause of the error you're seeing.

If you're running Docker on a Mac or Windows system and still seeing this error, it could be due to a misconfiguration or issue with your Docker installation. In this case, you may want to try reinstalling Docker or checking your Docker settings to ensure that host.docker.internal is correctly mapped to your host's IP address.

If you're still unable to solve your issues, you are welcome to contact the QuantConnect engineers by opening an issue in the QuantConnect/lean-cli repository on GitHub. Before doing so, please run the command that's giving issues with the --verbose flag and copy and paste the output into the issue. The --verbose flag enables debug messages to be printed, which makes it easier for the engineers to help you. This information is mentioned in the Troubleshooting section of the documentation.

I am running on a Mac.

Restarted Docker. Was able to RUN on the cloud. Still unable to DEBUG with BOTH:

localhost

and

I'm sorry to hear that you're still having trouble debugging with both localhost and host.docker.internal.

Given the information provided in the extracts, it's difficult to pinpoint the exact cause of the issue. However, it's worth noting that Docker-related issues can sometimes be resolved by checking Docker's file sharing settings. According to the Troubleshooting section of the documentation, you should ensure that Docker's Settings > Resources > File Sharing includes the necessary directories.

If you're still unable to resolve the issue, I recommend reaching out to the QuantConnect team for further assistance. You can do this by opening an issue in the QuantConnect/lean-cli repository on GitHub. Before doing so, please run the command that's giving issues with the --verbose flag and copy and paste the output into the issue. The --verbose flag enables debug messages to be printed, which makes it easier for the engineers to help you.

Restarted Docker. Was able to RUN on the cloud. Still unable to DEBUG with BOTH:

localhost

and

OK, I finally figured this out. In order to connect the VSCode debugger, one has to first:

Open a terminal in your organization workspace, run lean backtest "<projectName>" --debug ptvsd

via the CLI, as explained in: 

https://www.lean.io/docs/v2/lean-cli/backtesting/debugging