Arch Linux

Lockheed

Member

Registered: 2010-03-16

Posts: 1,550

How does kernel version influence filesystem?

In case of BTRFS, the newer the kernel the more up-to-date the btrfs code in it.

But I am confused how does it work in those cases:

case 1 - BTRFS filesystem is created on kernel 3.08. Kernel is later updated to 3.17 (and filesystem rebalanced if this is necessary(?)).
case 2 - BTRFS filesystem is created on kernel 3.17.

Is the filesystem from case1 worse? I.e. is it less modern, more buggy, and with fewer features than the one in case2?
Or does the kernel used for creating the filesystem not matter, and what matters is only the currently running kernel?

And, while we are at it, I wonder the same thing about influence of the btrfs-tools version.

lucke

Member

From: Poland

Registered: 2004-11-30

Posts: 4,019

Re: How does kernel version influence filesystem?

I'll post two examples of things that happen:

1) you create a filesystem with ext4, with default options from /etc/mke2fs.conf (part of e2fsprogs). A year later a new option appears in /etc/mke2fs.conf, newly created filesystems have it, your one-year-old filesystem doesn't - in some instances you could add it via tune2fs, sometimes the only way to get it is to make a new filesystem.

2) you create a btrfs filesystem on some kernel, some later kernel introduces a new mount option, you mount your filesystem with that new mount option on the new kernel, you are thus unable to mount it on older kernels.

Last edited by lucke (2014-11-06 09:54:28)

WorMzy

Administrator

From: Scotland

Registered: 2010-06-16

Posts: 13,576

Website

Re: How does kernel version influence filesystem?

A btrfs example is nodesize (formally known as leafsize). Previously, the default was 4k, but this was changed to a default of 16k a year or so ago. The only time you can set the nodesize is when the filesystem is created, so people on btrfs filesystems with 4k nodes needed* to recreate their filesystem(s) with the new defaults if they wanted to stick with what upstream recommends.

*note that this wasn't compulsory; old btrfs filesystems with 4k nodesize still work, and you can still create new btrfs filesystems with 4k nodesize.

Btrfs filesystems should remain compatible with all supported (mainline and LTS) kernels. So you shouldn't experience the latter part of Lucke's 2) scenario unless you're using an outdated kernel.

---

Sakura:-
Mobo: MSI MAG X570S TORPEDO MAX // Processor: AMD Ryzen 9 5950X @4.9GHz // GFX: AMD Radeon RX 5700 XT // RAM: 32GB (4x 8GB) Corsair DDR4 (@ 3000MHz) // Storage: 1x 3TB HDD, 6x 1TB SSD, 2x 120GB SSD, 1x 275GB M2 SSD

Making lemonade from lemons since 2015.