Bcachefs Lands Another Round Of Fixes For Linux 6.7

[LinAdmin]

Why are you using a broken filesystem? Doesn't make any sense to me.

I did not write broken, but broken design!

The only alternative FS with CRC is so much more clumsy that I preferred to invest an insane amount of testing efforts shooting randomly in only one of the Raid-1 partitions and then verifying that scrub did repair it.

After some time this looked stabilized, and as I already said that I cross fingers that it holds...

PS: This is my very last answer to an annoying discussion.

[PuckPoltergeist]

Ad b) I cannot believe that you did not suffer from the long period of erratic development of BtrFS Raid-1. You still can find my many postings on their websites. And BTW higher Raid-numbers today are not yet reliable.

Don't know about "their websites". Where can I read your many postings?

c) IMHO even today BtrFS is broken by design, although as said before, my servers run BtrFS Raid-1 without actual problems. I keep fingers crossed whenever updating it ...

Which way Btrfs is broken by design? I'd like to see the evidence. Why are you using a broken filesystem? Doesn't make any sense to me.

[LinAdmin]

a) I really need to add more sarcasm tags. I've never intended that Btrfs development was or is a nightmare.

b) If you think so, can you explain why? I've followed the Btrfs development from the beginning and use it since the early days. It was far from perfect but I wouldn't say nightmare.

c) What I meant initially is the behavior of Kent (and the fanboys/girls), the attitude that Btrfs is broken by design and his work (bcachefs) is so much better correct by design and only he knows how development is done and what is right. So I'm joking about it whenever another round of bcachefs fixes is announced.

Ad a) Those tags do not help to make your texts understandable.

Ad b) I cannot believe that you did not suffer from the long period of erratic development of BtrFS Raid-1. You still can find my many postings on their websites. And BTW higher Raid-numbers today are not yet reliable.

c) IMHO even today BtrFS is broken by design, although as said before, my servers run BtrFS Raid-1 without actual problems. I keep fingers crossed whenever updating it ...

There is no connection with bcacheFS and none of the participants should show fingers in the opposite direction.

I hope that within maybe a year bcacheFS Raid-1 will be reliable enough to replace my BtrFS Raids.

[PuckPoltergeist]

Ad a) Ok but rather special wording...

Ad b) In advance you never can predict if a design is truly reliable. There are many examples where the experts thought that it is perfectly reliable, and then tests or longer term use suddenly show ugly bugs...

Agreed that the development of BtrFS was and is a nightmare, but for the last few years Raid-1 with scrubbing works as expected. I cross my fingers and hope that BcacheFS in this configuration will be as good.

I really need to add more sarcasm tags. I've never intended that Btrfs development was or is a nightmare. If you think so, can you explain why? I've followed the Btrfs development from the beginning and use it since the early days. It was far from perfect but I wouldn't say nightmare.

What I meant initially is the behavior of Kent (and the fanboys/girls), the attitude that Btrfs is broken by design and his work (bcachefs) is so much better correct by design and only he knows how development is done and what is right. So I'm joking about it whenever another round of bcachefs fixes is announced.

[LinAdmin]

a) Native = upstream included, no external module
b) Reliable by design = in contrast to Btrfs it's designed right from the beginning, everything is well-conceived and perfected

Ad a) Ok but rather special wording...

Ad b) In advance you never can predict if a design is truly reliable. There are many examples where the experts thought that it is perfectly reliable, and then tests or longer term use suddenly show ugly bugs...

Agreed that the development of BtrFS was and is a nightmare, but for the last few years Raid-1 with scrubbing works as expected. I cross my fingers and hope that BcacheFS in this configuration will be as good.

[PuckPoltergeist]

What do you mean by "native" and "reliable by design"?
Anyhow, of what filesystem do you ramble?

Native = upstream included, no external module
Reliable by design = in contrast to Btrfs it's designed right from the beginning, everything is well-conceived and perfected

Oh, did I forgot sarcasm-tags?

[LinAdmin]

Unnecessary, the best native filesystem for Linux is robust and reliable by design.

What do you mean by "native" and "reliable by design"?
Anyhow, of what filesystem do you ramble?

[PuckPoltergeist]

>"... and the developers being aware of areas for (performance) improvements moving ahead"

IMHO highest priority should be robustness and the possibility of Scrubbing with automatic repair.

Unnecessary, the best native filesystem for Linux is robust and reliable by design.

[onlyLinuxLuvUBack]

At this point the priority should not be performance. To build a good reputation, the main thing is to focus on reliability, something that Btrfs never achieved after more than a decade. A high score in a performance test is of no use if it is not backed up by bombproof endurance. The most important thing about a file system, for the average user, is to forget that it exists.

It doesn't matter if it is at turtle speed but please: DO NOT CORROMPT OR LOSE MY DATA.

dude your data already corrupt ... "CORROMPT"

[alpha_one_x86]

crc32 -> maybe xxhash or similar will be better?

performance can imply disk change and/or huge code change (data risk), it's why do it in second time can be problematic. It's better do the changes now while nobody use the FS.