Those Using The XFS File-System Will Want To Avoid Linux 6.3 For Now

[phoronix]

Phoronix: Those Using The XFS File-System Will Want To Avoid Linux 6.3 For Now

Multiple users have been reporting metadata corruption issues on the XFS file-system when upgrading to the Linux 6.3 stable kernel...

https://www.phoronix.com/news/Linux-6.3-XFS-Metadata-Corrupt

[andyprough]

The Linux 6.4 kernel is reportedly working fine for some affected users

You know what else probably works fine? The 5.15 LTS kernel. The 5.10 LTS kernel. The 4.19 LTS kernel.

The frantic early adopters, using slower, regressed, buggier kernels for no reason related to new hardware - I do not understand their thinking.

[Leinad]

You know what else probably works fine? The 5.15 LTS kernel. The 5.10 LTS kernel. The 4.19 LTS kernel.

The frantic early adopters, using slower, regressed, buggier kernels for no reason related to new hardware - I do not understand their thinking.

I would be not so sure, if it is true expectation from article, that master works fine and fix was badly backported to older (stable) kernel

[mikus]

Or the poor bastards that use Arch and have to upgrade constantly lest they not be able to install things. I've been running 6.3.1 for a month or so via normal arch with xfs atop lvs/luks/mdadm, and no issues, though this is concerning.

[Britoid]

Or the poor bastards that use Arch and have to upgrade constantly lest they not be able to install things. I've been running 6.3.1 for a month or so via normal arch with xfs atop lvs/luks/mdadm, and no issues, though this is concerning.

Quite fortunately, in production XFS is generally only used on servers where it excels, and any decent IT admin will have those on a LTS distribution like RHEL-based distros which keep to lts kernels

[avis]

You know what else probably works fine? The 5.15 LTS kernel. The 5.10 LTS kernel. The 4.19 LTS kernel.

The frantic early adopters, using slower, regressed, buggier kernels for no reason related to new hardware - I do not understand their thinking.

Oh, boy.

• There's nothing indicating kernel releases on kernel.org are "unstable". In fact if you subscribe to LKML you'll find out that they are OK for general consumption.
• There's nothing indicating the LTS kernel releases that you're so proud of are "stable" as well. In fact on multiple occasions now, these kernels have seen serious regressions.

Maybe you could just admit that in general there's no guarantee that any Linux kernel is stable unless there's, you know, an organization standing behind it which does the necessary testing/regression work. Yeah, I'm talking about RHEL and Google which have fallen out of grace for the open source community. That's the only two kernels I fully trust. For everything else you're a perpetual beta tester, including and not limited to regarded as "stable" Ubuntu LTS and Debian.

And those "frantic early adopters", you're talking so kindly of, are saving your ass later on. Instead of, you know, thanking them and admitting that regression testing in general in Linux leaves a ton to be desired, you're denigrating them. The true face of the open source community!

This news piece has been brought to you by me, but Michael will never ever admit he's been tipped by someone. Not the first time. It's OK.

Have a nice day!

[andyprough]

And those "frantic early adopters", you're talking so kindly of, are saving your ass later on. Instead of, you know, thanking them and admitting that regression testing in general in Linux leaves a ton to be desired, you're denigrating them. The true face of the open source community!

Hey, as long as they all admit they are beta testers and guinea pigs for lab tests, and don't mind their data getting corrupted or working through undocumented bugs - I applaud them all.

And you are right, we are all standing on the shoulders of early-adopting neckbeard giants.

[Rallos Zek]

You know what else probably works fine? The 5.15 LTS kernel. The 5.10 LTS kernel. The 4.19 LTS kernel.

The frantic early adopters, using slower, regressed, buggier kernels for no reason related to new hardware - I do not understand their thinking.

It's a known fact that old kernels have old bugs and new kernel have new bugs. Check the changlog for an LTS kernel release and you will see patches for bugs that were already fixed months to years in the mainline and newish stable kernel.

[Brisse]

The affected Debian user mentioned in the article is either using a kernel from 'experimental' or spinning their own kernel. Debian testing and even unstable are currently on 6.1.27 and also frozen until the release of Bookworm so they should be unaffected, and are unlikely to ever pick up a kernel affected by this issue, so don't worry too much about it.

[JEBjames]

Michael

Grammar/wording

"Right now the Red Hat developers involved in maintaining the kernel builds for Fedora are waiting on more reports/testing for working to track down the problem."

should probably just remove "for working".