This seems to be kind of a real problem so let's state :

(This server is a home-made one)

• I self-host all my services with docker except for caddy
• Got four working HDD drives attached via SATA I think
• Works using ethernet and my pi-hole as dns
• Can not at all use a display

So, the issue I'm having is :

The server works for a few days, then randomly becomes unaccessible via ssh or any web port or anything really BUT continues seems to continue running cause ethernet green LED is still on et and "power on button" still on too.

After restarting, it usually resolves the problem but sometimes, it doesn't and I have to turn off the psu, wait a few hours or a day, and then it runs normally until it breaks again.

I'm not a linux professionnal so I'm reaching to you about how I could determine what's the problem ? Is there a journalctl log that could help me or anything that'd give me a hint ?

If I did not give crucial details tell me, Thanks for your help

Edit day 1:

(After restarting the server)

So after looking at htop (Yes I know I should've done it earlier) I got rid of two things that made my cpu be on 6 load average to 0.25 (was a dead immich_microservices container, a conflict between resolved and dnsmasq apparently AND a old matrix-synapse installation that .. pooped and loop-restarted).

I did a memtest on 3Gb by running memtester in the command line (after pluging a display I couldn't get to the grub option for real memtest god knows why) and no error reported, at least everything was "ok".

I am now ... waiting for the next goodbye of my server to then plug a screen and see.

I'll also try to get more RAM even though I got swap cause 4 Gb.... yeah

State of my little fucker right now :

https://preview.redd.it/yzjurxqz1cad1.png?width=1215&format=png&auto=webp&s=a38c4dea2f7f097f31db8d05c49a1a780c2e7593