Problem setting IP for host in HPE swarm learning

[DuyLe]

I can't set IP for host when setting up HPE swarm learning.
I'm setting up a connection between a Google Colab Jupyter and a GPU I rent from vast.ai. Can these 2 GPUs be connected? Or does my GPU have to be connected to the internet?
Hopefully, someone will reply to me soon. Thanks a lot!

[Suman_1978]

Hi,

Regarding setting up IP Address for host in HPE Swarm Learning, please refer to this guide and FAQ.

Regarding Google Colab, this may not be the right forum, but here are some references for you:
https://analyticsindiamag.com/explained-how-to-access-jupyterlab-on-google-colab/
https://colab.research.google.com/

Thank You!
I work with HPE but opinions expressed here are mine.
HPE Tech Tips videos on How To and Troubleshooting topics

[Vinky_99 · Accepted solution]

@DuyLe 
Hello,

To set the IP address for a host in HPE Swarm Learning, you need to modify the Docker configuration file. The file should be located in the /etc/docker/ directory and named daemon.json.

Here's an example of how to set a static IP address for a host in the Docker daemon configuration file:

{
  "bip": "192.168.0.1/24",
  "fixed-cidr": "192.168.0.0/24"
}

In this example, the bip option sets the IP address for the Docker bridge network, and the fixed-cidr option sets the subnet mask for the network. You can modify these options to suit your specific needs.

As for connecting two GPUs from different providers, it's possible as long as they are both connected to the internet and can communicate with each other over the network. You can use tools like SSH, VPN, or remote desktop software to establish a connection between the two GPUs.

However, keep in mind that connecting GPUs from different providers may introduce additional latency and network overhead, which can affect performance. It's also important to ensure that you have the necessary permissions and security measures in place to protect your data and infrastructure.

Hope this helps! Let me know