This worked great for me with Debian:
https://docs.zfsbootmenu.org/en/v2.3.x/
1 Like
1 and 2 are not good, but still not having reliable parity arrays after 15 years should be the nail in the coffin.
Forget about Btrfs. Pay attention to bcachefs, the Linux zeitgeist has moved on.
2 Likes
I thought I saw recent rumblings suggesting bcachefs was also falling out of favor, but canāt cite any at the moment.
1 Like
Linky linky
One example: https://www.phoronix.com/news/Linus-Torvalds-Bcachefs-Regrets
Itās funny that the in-tree thing is more disruptive to everyone involved than the āevil CDDL out of tree spawn of Satan Oracleā, despite the latter trying to be helpful with meaningful improvements to the Kernel that would benefit everyone.
3 Likes
Religious integrists and rational decisions rarely mix.
And the GPL is a cult.
RMS quite frankly said that the GPL is his means to destroy copyright be applying the rules of copyright.
The only āRMSā I know is a rock star! I play his hit single on blast over my car stereo system and crank it to the loudest volume!
Blessed are those who do not follow Ayatollah Richard M. Stallman, for theirs is the Kingdom of the BSD License.
(But donāt forget to work your statistical skills on Root Mean Square.)
4 Likes
Bcache promises so much and yet in my experience the only thing it did was to bork the file system. Then again, I keep having bad experiences with raspberry pi, so maybe itās less of bcacheFS and more raspi.
Are you implying that the GNU crowd recently came into possession of a lot of incredibly-cheap pagers and two-way radios?
I am of two minds on BCacheFS:
On the one side, a clean re-write with all the knowledge of preexisting file systems, volume managers and RAID schemes allows baking in features that are lacking in those preexisting solutions.
But, on the devils advocate side, their are plenty of problems. Lack of interest in helping BCacheFS because other solutions are both reasonably feature rich, (beyond what BCacheFS has today and for years to come). Plus, those other solutions have years or decades of use, making them much less prone to data loss. Another problem is single OS, (Linux), and tied to specific kernel release.
That last one, a file system tied to a specific Linux kernel release, is something that I have come to think is a bad idea. All non-kernel module Apps generally work across multiple Linux, (or other *nixes), kernels. And for small improvement change file systems, like EXT2 ā EXT3 and later to EXT4, it did not mater much that updates were tied to kernel releases. You simply did not use the newer EXT-x on OS loads for a while because it did not mater much. By the time it did mater, most Linux Distros had support and it was an easier choice to use the newer EXT-x.
However, with something as complex as BTRFS, BCacheFS or OpenZFS, maintaining the code out of Linux kernel tree seems to be a better choice, in my opinion. Especially if it is under active development like OpenZFS and BCacheFS. (Not sure BTRFS is under active development any moreā¦)
Of course having a major file system, and even one to be used with the root or ā/bootā file system, introduces complexities. But, for years OpenZFS has worked well on my 4 Linux home computers with at least monthly updates. (Some updates involved Grub, Kernel, OpenZFS or Initial RAM disks, an overwhelming majority worked fine.)
So, what do you think?
Are major file systems like OpenZFS or BCacheFS better off outside the kernel tree?
1 Like
General pupose file systems should, almost by definition, NOT be tied to a particular OS or a particular kernel.
In this story Torvalds acts as a better keeper of the Linux kernel than the FreeBSD maintainers did when they pulled some infamous commits about Wireguardā¦
But the lash about ānobody saneā using bcachefs for production begs the question what that experimental file system does in the kernel in the first place. Wilful testers should load it as a module or compile their custom kernel; experimental code should not be pushed to millions of production systems which will not, and by Torvaldsā own admission SHALL NOT, use it.
4 Likes
Another:
1 Like
Brodie Robertson is a fake Australian. At 00:08 he clearly says āZee FSā.
This video demonstrates how to properly say it in the Land of Winter July.
2 Likes
