search

LukeSmithxyz / landchad

Landchad.net Website Tutorial Page
The Unlicense
376 stars 87 forks source link

Suggestion: Approx platform size and RAM consumption #54

Closed craxed closed 2 years ago

craxed commented 3 years ago

Adding approx platform size and RAM consumption in each article under "Build your own platform!". Each platform have its own RAM usage and disk usage. For limited size servers, this would be helpful.

LukeSmithxyz commented 3 years ago

A good idea. People who host the listed services frequently should feel free to volunteer their footprints. For some things, I'm sure it varies a lot too.

ghost commented 3 years ago

I guess we could collect here, then figure out how to format and add it to articles. Gitea takes around 230MB for me. Miniflux (which I might write an article for in the future) takes 35MB, its postgres server 80MB.

To easily see ram usage per user (assuming you run services under separate users), you can use smem -uk, using PSS would be most fitting I suppose. Measuring RAM consumption is not as easy as one might think on Linux because of shared libraries (https://stackoverflow.com/questions/22372960/is-this-explanation-about-vss-rss-pss-uss-accurate).

termermc commented 3 years ago

Ergo (IRC server) takes around 10mb of RAM for me. This is with around 30 clients connected, maybe more.

‐‐‐‐‐‐‐ Original Message ‐‐‐‐‐‐‐ On Friday, July 2nd, 2021 at PM 5:47, phirecc @.***> wrote:

I guess we could collect here, then figure out how to format and add it to articles. Gitea takes around 230MB for me. Miniflux (which I might write an article for in the future) takes 35MB, its postgres server 80MB.

To easily see ram usage per user (assuming you run services under separate users), you can use smem -uk, using PSS would be most fitting I suppose. Measuring RAM consumption is not as easy as one might think on Linux because of shared libraries (https://stackoverflow.com/questions/22372960/is-this-explanation-about-vss-rss-pss-uss-accurate).

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or unsubscribe.

wdunkelheit commented 3 years ago

Pleroma can vary in memory usage based on how busy the instance is. But IIRC from playing with it in the past it never exceeded 300MB for a bare instance.

uranderu commented 3 years ago

I like the idea a lot. But sometimes it is a tough one because you are dealing with a lot of variable here :P ie Distro, Virtualization (Yes/no? Containers or fully-fledged VM's?), etc.

termermc commented 3 years ago

Assuming this is on a VPS from Vultr, it's a full VM using KVM, I believe. If you're talking about virtualization on the application side, according to the submission rules, none of the tutorials should be using virtualization/containers such as Docker, so that shouldn't factor into it at all. If things are running on a KVM instance like with Vultr, Linode, DigitalOcean, etc, it should behave normally. The only real variable on the server-side is how much swap is installed, but that's still not a huge problem, usually it's a gigabyte or two, almost never more than the amount of RAM. The amount of RAM applications themselves use will always been variable of course, but guidance on how much they typically use is pretty easy to establish on most things, provided that you explain your environment and use-case. If you are running a server and 5-10 people use it, most likely the RAM usage in such a use-case would be representative of what most people following the tutorial will be experiencing.

‐‐‐‐‐‐‐ Original Message ‐‐‐‐‐‐‐ On Tuesday, July 6th, 2021 at PM 10:02, uranderu @.***> wrote:

I like the idea a lot. But sometimes it is a tough one because you are dealing with a lot of variable here :P ie Distro, Virtualization (Yes/no? Containers or fully-fledged VM's?), etc.

— You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe.

ghost commented 3 years ago

If you are running a server and 5-10 people use it, most likely the RAM usage in such a use-case would be representative of what most people following the tutorial will be experiencing.

Well that's pretty much exactly my case. Here's what I got:

These all use up very little CPU most of the time except Synapse because it's soyware

NOTE: This is all on an Arch server with a pretty heavy amount of traffic (tards shill my stuff on le 4chinz), and quite a few users on my Matrix server. Will likely vary a bit for Debian servers but this is a general good guideline i guess