I think this was the right thing
I think this was the right thing
Posted Oct 2, 2025 16:22 UTC (Thu) by koverstreet (✭ supporter ✭, #4296)In reply to: I think this was the right thing by paulj
Parent article: Bcachefs removed from the mainline kernel
Excuse me? Follow the thread back to see where this started: bcachefs was called a "meme filesystem", and I was just explaining that there's a real demand for something better than btrfs.
_Every single time there's a bcachefs thread_ this happens. I start out explaining "yes, btrfs really does have issues, bcachefs really is delivering something better" and then people coming out swinging accusing me of attacking btrfs.
You're just trolling.
Posted Oct 2, 2025 16:26 UTC (Thu)
by jake (editor, #205)
[Link] (1 responses)
This sub-thread does not seem to be going anywhere useful for anyone. Can we please stop it here?
thanks,
jake
Posted Oct 2, 2025 17:06 UTC (Thu)
by koverstreet (✭ supporter ✭, #4296)
[Link]
Posted Oct 3, 2025 2:47 UTC (Fri)
by interalia (subscriber, #26615)
[Link] (3 responses)
I just want to say that that wasn't how I read it in context and I think you may have overreacted. The subthread was started by DemiMarie:
> I don’t think that experimental filesystems belong upstream. [snip] The need to get fixes out super quickly to recover user’s data does not mix well with the upstream kernel’s release cycle.
In reply, roryi said:
> It was very clearly marked as being experimental, though. I find it hard to understand who could possibly be using an experimental filesystem in a bleeding-edge kernel for important data without backups. [snipped]
> Has someone been actively recommending use of bcachefs to naive end users? Is there such a thing as a meme filesystem - and if so, has bcachefs somehow become one? Is there a rogue forum poster / youtuber / tiktoker out there tricking people into such risky behaviour without realising the implications?
I understood this reply to DemiMarie to be saying: "bcachefs was okay being upstream, because it was clearly marked as experimental. I would normally expect anyone using a filesystem marked as such in the most recent kernels would only be using it with backups, or with expendable data."
I don't think roryi was asserting that bcachefs IS a meme filesystem. In response to DemiMarie talking about data recovery they were asking if numerous users had somehow used a newer experimental filesystem without precautions, and IF so asking how they came to do so, finally asking speculatively if use of a new experimental filesystem somehow became a social media thing and thus encourage inexperienced users to do so. I thought it was a slightly odd thing to ask but I didn't consider it to be denigrating bcachefs's actual technical quality, so the entire subthread after that became an overreaction.
Posted Oct 3, 2025 3:29 UTC (Fri)
by koverstreet (✭ supporter ✭, #4296)
[Link] (2 responses)
Up until recently (~6 months ago?), it seemed I was perpetually telling the people who sounded less risk averse to "check back in six months"; this included people who had been bit by recent data loss on btrfs even when we'd hit the point where similar data loss was going to be unlikely on bcachefs (that's never been much of a concern on bcachefs, although we used to have a lot of "downtime event" bugs, until those were worked through).
There have been a few people who clearly ignored the experimental label and jumped in much too early and ended up unhappy and disgruntled, but - the people who ended up the most disgruntled are also the ones who like to tout their storage industry experience, so can't say I feel much for those guys :)
Generally speaking it's been a lot of younger hobbyists/tinkerers, or older people with spare machines who know exactly why we need this and have been investing significant amounts of time QAing and beating on it.
And for all the people I can talk because I've been interacting with, I should also mention there's a much larger userbase who hasn't been pushing all the various features quite as hard for whom it's been working just fine.
So there's a good mix. Some people are legitimately using bcachefs in production - less critical or less demanding scenarios - some are testing the limits, some people are putting it onto any old junk and running real workloads (!).
Especially in the last six months, it's gotten sufficiently solid that the "people running it on random junk with real production workloads crown" have been coming up with all kinds of stories of it quietly and happily humming along through dying hardware and god knows what else that people seem fairly confident would have rather upset btrfs or even ZFS.
So I don't think there's any reason for people to be taking special precautions at this point. The only special precaution you ever needed to take was "if it wedges itself, be patient and feed us logs/metadata dump so we can get you bugfixes", and even that's pretty much stopped and the bugs have become pretty mundane.
Posted Oct 3, 2025 4:18 UTC (Fri)
by interalia (subscriber, #26615)
[Link] (1 responses)
Rather than diverting my post into discussion of that, I'd be more interested in an acknowledgement that you might have misinterpreted what roryi said, or reasons why you think your interpretation was indeed correct if there was further context or subtext I didn't see.
Posted Oct 3, 2025 4:25 UTC (Fri)
by jake (editor, #205)
[Link]
This sub-thread is really not going anywhere very useful for anyone.
Please end it here.
thanks,
jake
Let's stop here
Let's stop here
I think this was the right thing
I think this was the right thing
I think this was the right thing
Please, stop
