Hello all,

Since installing the Seti@Home in Linux distributions, requires different versions of libraries. Does anyone created a Dockerfile,
so we can run it inside in a container ? Also supporting the Set@Home in the cloud?

Regards
Nikolaos Moraitis.

---

Hi Nikolaos,

I run BOINC in Docker on my Linux machines. The BOINC people publish official client images in the Docker Hub registry. These are a good place to start.

I build my own custom images because I need to support an older AMD GPU and because I prefer using Alpine base images. The official images generally come with the libraries needed for the more active BOINC projects like SETI, but you may need to extend them to add some libraries for certain projects. For example, I believe ClimatePrediction.net needs some additional 32-bit compatibility libraries if the containers run on a 64-bit system.

BOINC containers run just fine in the cloud, especially for processing CPU tasks. They're a bit trickier to configure for GPU tasks, but if you have a modern, well-supported graphics card, the documentation outlines everything needed to set the environment up. Just remember that, for GPU processing, we need to install the graphics card drivers on the host system as well as in the container (official image variants for each major GPU vendor include drivers).

After you get a container running, you'll need to connect the client to your SETI account. You can do this using the boinccmd command-line utility from the container (docker exec ...), or you can connect to the client remotely using the BOINC Manager GUI software. The official images consume environment variables used to configure the RPC socket for remote access to the client. For cloud deployments, be wary of exposing this port to the internet. The Docker daemon's support for TLS authentication can provide a more secure alternative to BOINC's built-in RPC protocol for remote access.