Re: EXT4 vs XFS

--nextPart2552078.NG923GbCHz
Content-Transfer-Encoding: 7Bit
Content-Type: text/plain; charset="utf-8"; protected-headers="v1"
From: Michael <[email protected]>
To: [email protected]
Reply-To: [email protected]
Subject: Re: [gentoo-user] EXT4 vs XFS
Date: Wed, 03 Jun 2026 15:52:21 +0100
Message-ID: <2626582.jE0xQCEvom@rogueboard>
In-Reply-To: <[email protected]>
References: <[email protected]>
MIME-Version: 1.0

On Wednesday, 3 June 2026 12:37:47 British Summer Time Dr Rainer Woitok 
wrote:
> Greetings,
> 
> sooner or later I will have to replace our two laptops.

If the replacement is planned in pursuit of higher performance AND the 
laptops are currently running spinning drives, then you may wish to 
replace their HDD for an SSD first and max out the RAM capacity by 
replacing/adding RAM modules - if RAM capacity is not already maxed out 
for the MoBo in question.  I have found the improvement achieved in 
terms of daily usage experience is sufficient to extend the useful life 
of an ageing laptop by a few more years.


> Mine is
> running Gentoo but sort of ages,  my significant other's  is running 
> Windows 10 (!)  and is even a bit older.  On the latter the
> requirements are roughly File Manager, Firefox, Office, VLC, Image
> and PDF Viewers, and that's about all.   So to reduce my maintenance
> effort I plan to install Gentoo on both new laptops.

The difference in speed/responsiveness between Win10/11 and any current 
Linux distro is significant.  Microsoft have managed to make Win10 
continuously slower over its lifetime and Win11 is not any better 
either.  Gentoo should give your partner's laptop a new lease of life.


> On my laptop I'm currently using EXT4 formatted disks and don't have
> any complaints.  Recently someone suggested XFS to me, and searching
> the Internet I found  mixed arguments  about which filesystem  to
> prefer.   Is anybody on this list using XFS  and would like  to share
> their experiences?
> 
> Sincerely,
>   Rainer

I had used XFS in its early Linux days.  Back then it would corrupt 
easily if the PC suffered a power loss.  This is not unreasonable, 
considering it was designed for use in datacenters with UPS, ECC RAM, 
and SCSI arrays.  I was using it on a laptop with an ever ageing 
battery. :p

After a couple of XFS corruptions and with its rather unreliable fsck 
utility I gave up on XFS and started using ext4.  I understand XFS has 
improved significantly since its early days with its metadata journal.  
Crash recovery problems have been overcome and automated corruption 
repair with the latest kernels is now possible.  Its strengths are the 
slightly higher speed, low fragmentation and huge scalability (delayed 
allocation to disk, use of extents and dynamic inode allocation).  Its 
probably the go to fs for an ever expanding media server.  It should 
work well on RAID because it can be tuned to align with striped array 
geometry.  However, it can't be shrunk without reformatting, does not 
have compression and it does not use native encryption.  I'm not sure it 
would benefit an OS root fs, but it could benefit a laptop partition in 
which you store large tar backup files, videos, etc.

I have used ext4 for a long time.  Multiple crashes, hard reboots, 
powercuts, you name it.  I even had a HDD failure or two.  I was always  
able to recover, perhaps losing the odd unsaved file in the process, but 
never the whole fs.  Unlike XFS you can shrink ext4 and you can encrypt 
it natively (fscrypt) in whole, or per directory.  It does not provide 
compression.

I suggest you rank and compare the fs features you need/want for your 
use cases, then you run a small test on the same PC and the same drive 
(NVMe/SSD/HDD) you intend to use them on.  This should give you real 
life experience, as opposed to benchmarks on the interwebs, all run with 
different parameters and hardware to yours and with different I/O 
bottlenecks.

--nextPart2552078.NG923GbCHz
Content-Type: application/pgp-signature; name="signature.asc"
Content-Description: This is a digitally signed message part.
Content-Transfer-Encoding: 7Bit

-----BEGIN PGP SIGNATURE-----

iQJPBAABCAA5FiEEXqhvaVh2ERicA8Ceseqq9sKVZxkFAmogP6UbFIAAAAAABAAO
bWFudTIsMi41KzEuMTIsMiwyAAoJELHqqvbClWcZ3QAP/A5BNMd7JasYllXr+MpW
N1tlWrhvmIGzmno96RfJTYXns/TfHA35F5Q3CBGrAGegrjdi83fnCGG++OuTYz3p
EL1F46Etmm1kRI3gVrjn3BdcJdobvjqhaYBTYEV7JyixBZO01aG+6ISFm+QHTT3X
BEKAoG+7ligRac4En68dpZcaIErLBcosq7znZnxwvVn8wPNkIfqmYQb7y0/KKLn5
vk2494eynI7YrxkRuoUirzj9A1Q700aWrXLDgEX1EqmHIUhDnzPbb7YYJYmsyN38
pNPrA3ovn2kG3mGt9ZhuHveDqhm8xerxJIQWj7O95gmXPnf7YtAeWfNBFptX1Vwg
NUURNwifmLJ7a6PUCPqZYBiGprLn0ejMUUmFwxmpZ2JZ5h81S0NFk3SLzGBHTk51
m+iWkRJxf2zrvWAUSTTeX9PFCxn8XhIjAQE/PmQGy7E0X+8qpvH8XNtxaEk7bCy6
+1ymQDHigpm9FJcKQCMg3LK49R2xv0O0xmHFtJV4B35zpu8LJn8PvveoAPF0kh35
RDlpccGOLwVsow9dsnquXJNKh7D8PJMccAO/DCIDtn2ZA7wcWaHmANphy1xHCUns
zuIKIdK3rPXDGN3oX3AAR1OIQDwAwST2AgTU4ulby3+/Cyu9vgrLabV1tE/zC88A
7I0//PSa37EKmpu/m5yWHWpa
=c6GC
-----END PGP SIGNATURE-----

--nextPart2552078.NG923GbCHz--