BOINC - Unraid Setup/Multiple Instance Guide

[nadoteg]

I know a few of us in the BOINC community have instances on Unraid, so I wanted to throw a guide together to help with initial setup and adding other instances, similar to this guide for windows: https://www.overclock.net/threads/guide-setting-up-multiple-boinc-instances.1628924/

 

Unfortunately, I only use NVIDIA GPUs so I don't know what changes need to be made for AMD or Intel GPUs. I would imagine they are similar but I am going to assume for this setup that you are using an NVIDIA GPU.

Initial Setup
The prerequisites for this guide are:

• You already have docker set up on your machine and community apps activated. There are plenty of guides to do this part so I won't include this here.
• If you would like to use an NVIDIA GPU, you also need the NVIDIA-driver plugin, which you can get from the community apps.

This uses the official BOINC docker image (https://hub.docker.com/r/boinc/client). The default docker image tag (boinc/client) should work, but be sure to check out the other tags (boinc/client:amd, boinc/client:nvidia, etc.) if things aren't working the way you expect: https://hub.docker.com/r/boinc/client#supported-architectures-and-tags

Add the community app:
Start by going to the apps tab, and searching for "boinc", and clicking "install" on the official container:

You will be presented with some configuration parameters. I'll go through the important ones:

 

1. This image should be fine, but like I said in the beginning, if you have issues or want to try things out, try a different tag that aligns with your system or interest: https://hub.docker.com/r/boinc/client#supported-architectures-and-tags
2. By default this is "Host" but set this to "Bridge" so we can use this as a template for more instances.
3. The default appdata path is fine.
4. Set this to whatever you like, be sure to remember it, because this the password for managing it with another computer.
5. Leave this as is. 
6. We will need to add some additional configuration parameters in the next step. Click this to add one.

Set additional configuration parameters:

 

There are a few configurations we need to add so I'll use the numbers painted in. 

NVIDIA GPU users only. Skip if you are only using CPU or have another GPU.

1. Variable
2. NVIDIA GPU
3. NVIDIA_VISIBLE_DEVICES
4. all

NVIDIA GPU users only. Skip if you are only using CPU or have another GPU.

1. Variable
2. NVIDIA Capabilities
3. NVIDIA_DRIVER_CAPABLITIES
4. all

 

Required for template, so we're going to include this here.

1. Port
2. RPC
3. 31416
4. 31416
5. TCP

Nvidia runtime and CPU Pinning (For NVIDIA GPU)
Lastly, at the top right, click on the toggle by "Basic View" so it turns on "Advanced View". You should see a few more options. By "Extra Parameters" type in "--runtime=nvidia" without quotes. A little further down, you can select which threads are used for this container. This is useful for partitioning off a thread for GPU usage.

Once you have these set up you can select "Apply" and the container will install and start. When adding it to the BOINC manager, use the IP of your unraid machine with the "host port" you entered in the additional configuration parameters. When adding a new BOINC manager window from the menu use <ip address>:<port>.

 

Double check that this works before moving onto the next step, or if this is the only one you want, you are finished.

Additional Container Setup
To set up another container for bunkering or GPU usage, go to the bottom of your Docker page, and click "Add Container". On the next page select the template dropdown, and scroll down until you see the name of the BOINC container you just made under "user templates". We'll want to incrementally change a few things, but we should not have to do much since we set everything up in the template.

 

• change the name of the container from "boinc" to "boinc2" or whatever is next. This must not be the same as another container.
• change the appdata folder from "boinc" to "boinc2" or whatever is next. This must not be the same as another container.
• change the RPC host port from 31416 to 31418 or whatever is next. This must not be the same as another container. Do not change the container port part of the parameter.
• adjust your CPU pinning (optional)
• change the password (optional)

 

Once you have all those changed, you can hit Apply and your new container will be added. Repeat for additional containers.