Describing an Incus environment on research papers

Incus
1

I’m writing a paper with some experiments running in an incus cluster. It is customary to describe the environment for running experiments and it’s the first time I’ve run them using incus, which changes things around.

Currently it’s written as follows.

Captura de tela de 2025-02-28 18-48-49
Captura de tela de 2025-02-28 18-48-49739×197 56 KB

Any comments or improvements on this?

Some details which I’m not sure about, for example. Should incus be capitalized? I’ve called the lxc containers as Incus lxc containers just to make sure they’re managed by incus. Should I just call them lxc containers? Does the incus version specify the lxc version? Should it matter?

2

A few notes, though it’s mostly nitpicking :slight_smile:

  • Amount of memory is usually in GiB rather than GB (at least physical memory is typically in GiB, you may have actually used an amount in GB for your VMs)
  • lxc is usually capitalized, so LXC
  • You can’t infer the LXC version from the Incus version but it’s not particularly important and won’t affect performance

For anyone wanting to reproduce the environment, the most important tends to be the hardware it’s run on, the kernel version, the storage driver (if I/O matters) and then the information on the instances (containers vs VMs, how much CPU allocated, how much memory allocated).

Might be worth mentioning if the machines are running headless (server) or are using a desktop installation. Given the CPUs in use here are consumer grade desktop chips, the distinction is important to know if the systems are clean minimal server installs or if they’re busy with also running a desktop.

3

I don’t know what the convention is, but I generally say someone is running “on” an architecture, “on” a cluster, etc.

You use are using the word “in”.

4

The way I read it, he was describing the physical machines themselves being in an Incus cluster which sounds right to me.

I agree that when referring to the instances themselves, they would then be running on a cluster.

(Note that it’s 3am here so my eyes may be playing tricks on me ;))

5

Good catch. That is true for one of the three uses of in.

I am still drinking my morning coffee and my eyes played a trick on me. :dizzy_face::coffee:

6

Thanks a lot for the replies. I was mostly looking on how to mention Incus and LXC properly, and this certainly helped.

Indeed this is related to reproducibility, but usually not enough detail is provided. Historically people mention just CPU, OS, kernel version and amount of RAM. It helps to provide all source code, and it’s usually reproducible on different architectures, so it’s not usually a problem.

Just some comments on the help provided, it is actually GB in this case, at least it’s what I’m using for the incus limits, since if fits better into the total RAM we have. I will just have a quick check on RAM later. LXC version and where the incus packages come from will probably not really matter and should be reproducible on similar environments, so I’ll not deviate from the norm of not providing too many details.

I’ll also check the proper use of English, if I can. But these are usually caught in the editorial process.

Anyways, thanks a lot! I really like the whole incus project and wanted to make sure I’m not mentioning it incorrectly in any way.

Edit: Mentioning headless is nice and just a single word.

1 Like