Adamantite

Adamantite is a current work in progress.

What it will do

Adamantite is currently a command line tool similar to htop and top. It will provide a digestable overlook on how much system resources is being consumed by Hytale and its processes.

Why?

My reasoning for building this tool is because I wanted to learn rust. But also I host my own hytale server and at one point my server was lagging and attempting to use htop to figure out why was difficult.So I thought I wish there was a way that I could get a high level overview on how Hytale and its processes are performing on your system.

Usage

Installation methods

There is binary realeses for linux, you can clone this repo, or run

cargo install adamantite

Track a system resource over 5 seconds and output as a csv.

Setting up adamantite configuration file

Adamantite searches for pid of hytale via systemd. If your hytale server is running as a service, you can configure adamantite to search for that unit name.

To set your unit name for your service, lets say "hytaleserver" run this:

adamantite config unit-name "hytaleserver"

This will update adamantite config.toml located, on a linux system at least, at ~/.config/adamantite/config.toml

To track a cpu over 5 seconds and output as a csv run:

adamantite track cpu -t 5 -o csv

There are two system resources that you can monitor for x seconds:

• cpu
• mem

These are cpu, memory.

Output example:

This command will return to you the average number of cores being used by hytale. Adamantite will also provide the average cpu cores your system is using so you have a refrence on how hytale is performing compared to the rest of your system.

This command will lastly provide the average cpu load your entire system experineced over x seconds.

Example command:

$ adamantite track mem -t 5

Output would look like this

This command will provide the average memory usage in gigabytes over x seconds. It will also provide the average memory usage in gigabytes as well for the entire system over x seconds.

For the rest of the output, adamantite will provide average memory usage as a percentage for hytale, your entire system including hytale, and your entire system excluding hytale as well.

Example command:

$ adamantite track mem -t 10 -o csv

This command will write to a file with either the name mem_usage.csv or cpu_usage.csv based on the resource that is currently being tracked.

An example of the file contents are this

To see io pressure that your system is currently experiencing:

adamantite pressure io

Shows how often system work is stalled due to resource contention

You can view the past pressure for two types: io and mem. Memory pressure is time stalled because memory could not be allocated or reclaimed fast enough. I/O pressure is time spent waiting on disk or storage operations to complete.

Why is pressure important

Pressure directly reflects time the server could not advance its game loop, making it a strong indicator of tick instability and perceived lag.

To view live feed like htop:

$ adamantite live

This command will bring up a TUI where you can see live updates of your system memory and cpu usage as well as hytale's memory and cpu usage.

TUI will now also present hytale logs when they happen. You can scroll through them using j for up navigation and k for down navigation.

Currently investigating feature to add

Currently looking into providing disk metrics that should be useful in determining perfomance drops for you server

What is being worked and plans for future

Maybe might be possible to use previous runs of commands and compare them with other runs. Say you ran the adamantite to track hytale's memory usage over 5 seconds while the system was idle. Now say you want to compare that when you have three people in your server. I was thinking you provide two outputs from the command and compare them.Might be cool to see how hytale performance differs.

Say we track 5 seconds of cpu usage from Hytale and its processes. We need to follow Hytale's logs that it produces at that same time.

There are two possible routes and probably more to do this but these are what came to my mind first.

We can go the concurrency route where we instantiate two threads, one thread to get Hytales cpu usage and another thread to parse Hytale logs.

If not concurrency, I looked into using journalctl. So once we get the cpu usage for those 5 seconds we just pass arguments to journalctl to show me Hytale's logs for the last 5 seconds and parse it from there

AI ?

Currently no AI generated code. Goal for this project is to provide code that I understand to you as a user.