Question 1

Documentation

Accepted Answer

Official Documentation

For Vast.ai official documentation, go to https://vast.ai/docs.

If you are just getting started, read our QuickStart Guide.

CLI documentation

Vast.ai has a command line interface written in Python. The Python script calls our API. For the latest documentation on the CLI, use the --help command.

To install the CLI, you will need to authorize your API key. Our documentation in the product uses pip. The Python script has it's own GitHub repo.

API Documentation

The complete API documentation is located below the CLI documentation on the CLI docs page. You can see the indivudal functions available on the left hand navigation.

Question 2

Alternative method to back up data using Rclone

Accepted Answer

What is Rclone?

Rclone is a command-line program to manage files on cloud storage. It is a feature-rich alternative to cloud vendors' web storage interfaces. Over 70 cloud storage products support rclone, including S3 object stores, business & consumer file storage services, and standard transfer protocols. Official website: https://rclone.org/

Why Rclone?

The following links will take you to Vast.ai, which already offers official documents for copying from and to multiple cloud storages: https://docs.vast.ai/documentation/instances/storage/cloud-sync https://docs.vast.ai/documentation/instances/storage/cloud-backups

This article discusses the benefits of using Rclone as an alternate data backup method.

How can we utilize Rclone in Vast.ai instances?

The majority of Vast.ai's templates already have rclones installed. Open Jupyter Terminal and enter "rclone version" to confirm the version.

(main) root@C.xxxxxxx:/workspace$ rclone version
rclone v1.60.1-DEV
- os/version: ubuntu 24.04 (64 bit)
- os/kernel: 6.8.0-59-generic (x86_64)
- os/type: linux
- os/arch: amd64
- go/version: go1.22.2
- go/linking: dynamic
- go/tags: none

For some cloud storages, the old version of Rclone may not work, so it is suggested to upgrade rclone to the latest version using this command: sudo -v ; curl https://rclone.org/install.sh | sudo bash

``` (main) root@C.xxxxxxxx:/workspace$ sudo -v ; curl https://rclone.org/

Question 3

Feasibility of Machine Hosting on WSL

Accepted Answer

Why You Can't Host on Vast.ai Using WSL: Technical Constraints and GPU Virtualization Limits

Q: Documentation

Official Documentation For Vast.ai official documentation, go to <a href="https://vast.ai/docs">https://vast.ai/docs.</a> If you are just getting started, read our <a href="https://vast.ai/docs/overview/quickstart">QuickStart Guide</a>. CLI documentation Vast.ai has a command line interface written in Python. The Python script calls our API. For the latest documentation on the CLI, use the --help command. To install the CLI, you will need to authorize your API key. Our <a href="https://cloud.vast.ai/cli/">documentation </a>in the product uses pip. The Python script has it's own <a href="https://github.com/vast-ai/vast-python">GitHub repo</a>. API Documentation The complete API documentation is located below the CLI documentation on the <a href="https://vast.ai/docs/cli/quickstart">CLI docs page</a>. You can see the indivudal functions available on the left hand navigation.

If you are looking to monetize your GPU hardware on Vast.ai, you might be tempted to use Windows Subsystem for Linux (WSL2) to avoid leaving the Windows ecosystem. However, according to the official Vast.ai hosting requirements, a native Ubuntu installation is mandatory. This isn't just a "recommendation"—it is a result of fundamental technical barriers within the WSL2 architecture, particularly regarding how it handles GPU resources and low-level system hooks.

${color}[#000000](1. The Virtualization Barrier: Para-virtualization vs. Native Access)

WSL2 operates on a highly optimized version of the Hyper-V utility VM. While it provides a Linux kernel, it uses GPU Para-virtualization (GPU-PV). In a native Ubuntu setup, the Nvidia driver communicates directly with the hardware. In WSL2, the "driver" inside Linux is actually a projection of the Windows host driver (dxgkrnl).

The Problem: Vast.ai’s hosting daemon requires NVML (Nvidia Management Library) to have direct, bare-metal access to the GPU to monitor temperatures, power limits, and clock speeds.
The Limitation: Under WSL, many NVML functions are restricted or report inaccurate data because the hardware is being managed by the Windows Windows Display Driver Model (WDDM), not the Linux native kernel.

${color}[#000000](2. Docker-in-Docker and Cgroup Constraints)

Vast.ai functions by running customer w

Question 4

Instance is in scheduling status, but the same machine is available for renting

Accepted Answer

Observed Behavior:

The instance is in scheduling status
However, the machine is still shown as Available for rent on the portal (Machine ID: 37485).

Explanation:

When an instance is created, it is bound to a specific GPU ID. If the instance is stopped, it remains bound to the same GPU ID and waits for that GPU to become available again.
On machines with more than 1 GPU, other GPUs may still be free and available for allocation. As a result, the same machine can still appear as available on the portal, even though the original instance is stuck in Scheduling while waiting for its assigned GPU.

Suggestion:

We recommend using volume storage to persist data and enable access across multiple instances on the same host machine. If an instance gets stuck in Scheduling but other GPUs on the same host are available, you can create a new instance on that host and mount the same volume to continue working on it.

Question 5

Hosts cannot search for their machines

Accepted Answer

Issue

When a host searches for their own machine, the machine doesn't appear in the list (both CLI and portal)

However, when searching, other users can see the machine.

Root cause

The machine has an active instance of the host account. To get the machine to show up in the search list, that instance needs to be destroyed.

Question 6

Missing GPUs earning

Accepted Answer

Issue

The machines have instances running

But there is no earning for those machines

Root causes

One possible reason that the host has instances running on those machines by themselves and there is no earning for those instances

Featured & Frequently Read Articles

Browse All Categories