Edit to clarify: You’re just explaining back-end stuff that should be completely invisible to users (and normally is). The parent comment specifically mentioned partitions, when you install a new Linux OS the installer asks you “how do you want your drive split up? where do you want the swap, and how much?” etc etc. which a newbie can’t even begin to answer, it shouldn’t even ASK that if the user didn’t specifically choose to set this completely manually.
As another commentor said most Linux distros will do that for you.
If you still aren’t sure then the answer is generally: swap should be at least as big as you have RAM at least if you want to hibernate. If you don’t want to hibernate then you can make it smaller but it might impact system performance when low on memory.
For file systems they will often offer you LVM or ZFS or occasionally BTRFS as options. These all allow you to make system snapshots (like the concept of restore points in Windows). If you like the idea of that then say yes, otherwise you will get slightly better performance not using those systems. ZFS and BTRFS also have other uses like RAID-like functionality and detecting and possibly correcting data corruption - with only one drive these features are not as useful though.
BCacheFS is new to the mainline kernel and does much the same as ZFS and BTRFS, when distros start offering this as a supported option it’s probably a good idea to use this, kind of unfinished at the moment though.
Edit: also if it asks if you want a separate /home generally you want to say no. Unless it’s a btrfs subvolume or zfs dataset in which case say yes.
ZFS and BTRFS also have other uses like RAID-like functionality and detecting and possibly correcting data corruption - with only one drive these features are not as useful though.
Yes they are. Regarding data integrity (bit rot or read/write errors) though. Regarding data backup, no.
BCacheFS is new to the mainline kernel and does much the same as ZFS and BTRFS, when distros start offering this as a supported option it’s probably a good idea to use this, kind of unfinished at the moment though.
I think kernel support is planned for version 6.7… I think.
Yeah, that is still too unsable in real world scenarios.
Still, it’s great that people are doing this from scratch, not just build on what Oracle throws at us. I do wish they address the RAID5/6 issue properly, not just throw it under the carpet like BTRFS. I did hear that they are working on fixing RAID5/6 in BTRFS, but I haven’t looked at how far things are. I still use it with mdadm in the background for the array.
To be fair, I haven’t thought about the USB low-levei devices in years. Nowadays most things just work, and if they don’t you do a quick lsusb to check it actually sees the device then you Google what package you need to install.
Yeah… compiling from source to get a kernel module loaded is a part of the past now. Though we still do it from time to time, for some obscure pieces of hardware (like my Microtek scanner 😒), but things just generally work out of the box.
It’s not exactly backend. It’s how UNIX like OSes work. Sooner or later, you’re gonna have to learn this. The idea behind what @[email protected] said was (IMO) better have this in the back of your mind when it comes to Linux, cuz you’re gonna need it sooner or later. Just keep it tucked away for when the time comes.
this doesn’t help at all
Edit to clarify: You’re just explaining back-end stuff that should be completely invisible to users (and normally is). The parent comment specifically mentioned partitions, when you install a new Linux OS the installer asks you “how do you want your drive split up? where do you want the swap, and how much?” etc etc. which a newbie can’t even begin to answer, it shouldn’t even ASK that if the user didn’t specifically choose to set this completely manually.
most installers for just works distros give you a recommended configuration that you can just click “yes” to
As another commentor said most Linux distros will do that for you.
If you still aren’t sure then the answer is generally: swap should be at least as big as you have RAM at least if you want to hibernate. If you don’t want to hibernate then you can make it smaller but it might impact system performance when low on memory.
For file systems they will often offer you LVM or ZFS or occasionally BTRFS as options. These all allow you to make system snapshots (like the concept of restore points in Windows). If you like the idea of that then say yes, otherwise you will get slightly better performance not using those systems. ZFS and BTRFS also have other uses like RAID-like functionality and detecting and possibly correcting data corruption - with only one drive these features are not as useful though.
BCacheFS is new to the mainline kernel and does much the same as ZFS and BTRFS, when distros start offering this as a supported option it’s probably a good idea to use this, kind of unfinished at the moment though.
Edit: also if it asks if you want a separate /home generally you want to say no. Unless it’s a btrfs subvolume or zfs dataset in which case say yes.
LVM has snapshot support 🤨? Is this a new thing?
Yes they are. Regarding data integrity (bit rot or read/write errors) though. Regarding data backup, no.
I think kernel support is planned for version 6.7… I think.
Yeah it’s included in kernel 6.7. Still probably too new to really use as they are still working on certain features like erasure code/parity.
Without another drive it can’t rebuild data, so all it can do is detect corruption not correct it.
Yeah, that is still too unsable in real world scenarios.
Still, it’s great that people are doing this from scratch, not just build on what Oracle throws at us. I do wish they address the RAID5/6 issue properly, not just throw it under the carpet like BTRFS. I did hear that they are working on fixing RAID5/6 in BTRFS, but I haven’t looked at how far things are. I still use it with mdadm in the background for the array.
I would just stick to the defaults.
Systems don’t always give defaults. That’s what the guy I am replying to is complaining about.
To be fair, I haven’t thought about the USB low-levei devices in years. Nowadays most things just work, and if they don’t you do a quick
lsusbto check it actually sees the device then you Google what package you need to install.Yeah… compiling from source to get a kernel module loaded is a part of the past now. Though we still do it from time to time, for some obscure pieces of hardware (like my Microtek scanner 😒), but things just generally work out of the box.
The Linux mint installer does this automatically. All you need to choose is whether you want to wipe your drive or install along side a existing os
Edit: added specifics
It’s not exactly backend. It’s how UNIX like OSes work. Sooner or later, you’re gonna have to learn this. The idea behind what @[email protected] said was (IMO) better have this in the back of your mind when it comes to Linux, cuz you’re gonna need it sooner or later. Just keep it tucked away for when the time comes.