speedtest-fileserver-rs

This is a server that serves random data. It accepts URLs of the form http://domain.name/<size><multiplier>.<ext> .

For example, when running on localhost port 3000:

http://localhost:3000/10MB.bin
http://localhost:3000/2GiB.bin
http://localhost:3000/97KiB.bin

.. and it will serve a file of the requested size consisting of random data.

The directory index http://domain.name/ serves a dirlisting of a number of files with common sizes in the range of 1MB to 10GB.

Features.

• fast.
• serves completely random data.
• index file can be customized (handlebars template).
• http and https support.
• can write access log files.
• written in Rust without any unsafe code.

Building it.

First install rust, if you haven't yet. Note that you need a recent version of Rust. The version that comes with your OS might be too old.

Run this (as yourself or a development user, not as root):

curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh

Then clone the repository, cd into it, and build the server:

then build the server with

git clone https://github.com/XS4ALL/speedtest-fileserver-rs
cd speedtest-fileserver-rs
cargo build --release

The binary will be present in target/release/speedtest-fileserver-rs.

There is an example configuration file provided, you need to install that in /etc or tell the server where it lives with the --config command line option.

Building .rpm or .deb packages:

• building a debian package
• building a rpm

Configuration.

The features mentioned above can be configured via the configuration file. See the comments in the example configuration file.

Bugs.

When access-log logging is enabled, the server logs download size and speed (seconds elapsed, with ms accuracy). However due to internal buffering in the libraries used and the OS itself, a logentry is written before all data has been transferred to the client. The buffering might be a few MB, so using size-of-download and time-elapsed to calculate download speeds is not recommended, unless the download size is considerably larger than the size of the internal buffers (say, 1GB and up).

If you need precision in the logs, try if a front-end proxy like nginx has better accuracy.