Very often people who self-host don't have proper storage for their system. Leaving the server in several parts, in a high traffic area, accessible to children or pets, or in a poorly ventilated area can quickly turn into a disaster.
Ideally, your hard disks should be fixed to avoid vibrations which can accelerate the wear of the equipment or even reduce its performance, especially if there is another disk next to it.
If you use a swap file with an SSD or SD card with too much swapiness, your storage media could give up the ghost prematurely due to too many writes.
To avoid this:
If it is above 10:
sysctl vm.swappiness=10 nano /etc/sysctl.conf
If present, change the vm.swappiness value to 10. Otherwise add the line:
vm.swappiness = 10
In order to limit hardware failures of storage media, it can be relevant to set up a cluster of mirrored disks (RAID, ZFS). The idea here is that everything that is written to one disk will be written to the other. This way, if one fails, the other continues to work and the server is still fully functional.
There are also more advanced clusters that maximize fault tolerance (failure of 2 disks like RAID6) or storage (see RAID 5).
However, these disk clustering techniques should not be considered as backups. A RAID array should be considered as a single storage medium. Indeed, if this technique avoids having to reinstall in case of a probable disk crash, it is far from zero risk.
Some examples of situations known to professional system administrators:
If you want to set up a RAID array or use btrfs, the easiest way is to do it at installation with the YunoHost iso in expert mode (when partitioning the system).
Found errors? Think you can improve this documentation? Simply click the Edit link at the top of the page, and then the icon on Github to suggest changes.
Powered by Grav + with by Trilby Media.