Bcachefs Looks Like It Won't Make It For Linux 6.6

[Old Grouch]

I am German and I we don't send 5000 sentences garbage before we come to a point.

For an American writer's perspective on the German language, I recommend these three essays by Mark Twain.

[ Für die deutsche Sprache eines amerikanischen Schriftstellers Aussicht, ich von Mark Twain diese drei Essays empfehle . // For, the German language, an American writer, his perspective, I, of Mark Twain, these three essays, recommend. ]

(For an academic background on one of the essays: "The Awful German Language", I suggest reading the article by Norbert Hedderich of the University of Rhode Island doi:10.2307/3531681.)

[avis]

so what is the solution and how to apply it? (also previous kernels appear to be affected)

Download more RAM.

[MorrisS.]

Download more RAM.

What this means? Windows operating systems aren't affected by the same issue. RAM cannot be downloaded. Do you mean that since linux has a deficit, user has to buy more RAM? It appears a bad solution.

[avis]

What this means? Windows operating systems aren't affected by the same issue. RAM cannot be downloaded. Do you mean that since linux has a deficit, user has to buy more RAM? It appears a bad solution.

That was sarcasm, yes, installing more RAM and disabling SWAP altogether (I've been running swapless for over 20 years now) will fix the issue. Running without SWAP in absolute most cases is more preferrable than with SWAP. If you read anything which contradicts with this statement on the net, it's just plain wrong. SWAP is only necessary when you don't have enough physical RAM to accommodate all the running applications and if that's the case, latency will be horrible. Yeah, Linux uses SWAP for hibernation but it's a brain damaged concept and a bad idea which should have never been implemented this way. Windows has a separate hibernate.sys file for that (or something like that - I don't remember its name).

[timofonic]

Kent Overstreet replied Linus Torvalds, so there's some hope...

https://lore.kernel.org/lkml/[email protected]/

[reba]

What this means? Windows operating systems aren't affected by the same issue. RAM cannot be downloaded. Do you mean that since linux has a deficit, user has to buy more RAM? It appears a bad solution.

Several options (from best to worst):

1) buy and install more RAM, it's not too expensive and helps in every situation (fs caches, more breathing room for applications, etc)

2) use zramswap (Debian: package "zram-tools") to create a RAM-based swap partition. Sounds weird but the memory for this swap is compressed, so it acts like kind of a very very quick memory compressor (remember the tools from the 90s?). Honestly, it works really well and I use it on my system (lz4, 50%) as an emergency backup in case an application runs amok as it gives me time to react and kill it before the system starves of RAM and the OOM-killer is activated or when running several virtual machines in parallel and doing stuff. Cheap, easy, effective, not damaging.

3) configure the system to be less "swappy", means to not use the swap too early. Gives more RAM-pressure to applications and harms cache

4) disable swap altogether, let the system handle memory allocations within the constraints

5) put the swap on SSD (as partition to reserve the space or file to allow growing/shrinking), it is very quick (but really slow compared to real RAM) and hurts the health of the SSD with its frequent writes. Wait, this is actually a bad advice

Never swap to HDD, it's just too slow nowadays as system and programs gobble up RAM much faster then the transfer to physical swap can cope with it nowadays.

[fitzie]

After reading everything, I think everyone here is somewhat right about their opinions.

Linux kernel management has to substantially improve, it's extremely incomplete. Documentation is incomplete, not every procedure is completely documented. There's massive holes in documentation that causes lots of confusion and conflicts. There should be full-time paid managers to cope with certain tasks, such as training and conflict solving.

Where's Andrew Morton, Greg Kroah-Hartman and others here? Can anyone please provide a proper organization diagram with the duties of them? I'm curious about it. Any URLs about it other than the so 80s MAINTAINERS file?

It's extremely easy to call people passive aggressive, and impugn their motives, but I just don't see it having an outsized effect in this circumstance. Certainly I don't think we need a revolt like the CoC thing against Linus a few years back, (and the CoC was not useful at all here, and it was only mentioned during these events briefly and dismissed).

The linux kernel project is extremely mature and extremely unique. That's not to say it's perfect or there's not simple things that they should and can do, but there's limitations on what can be done. Linus has said in the past the roadmap of Linux is what people want to work on and submit. That's pretty much it.

I don't think putting more burden on maintainers and lieutenants to handhold anyone trying to submit code is without cost. Linus has said before they are in that role because they have the quality of having "good taste". That doesn't mean they are optimized for interpersonal skills.

Both Kent and this forum made quite a big deal of these recent events, so we lack perspective. Time will tell if something will change or come of this.

As for the Linux project, there's really just Linus. There's quite a bit of project support at LF, but not really for managing kernel development directly.

The people who tend to get involved in kernel community stuff include James Bottemley and Ted Ts'o (who got involved but their efforts were fruitless). It's really a massive space, where these senior people tend to not have fixed responsibilities or roles for long time. Morton was like junior Linus way back when but these days he's really just MM stuff. Rik van Riel created kernel newbies but these days he keeps his head down as well.

I think this is all going to blow over. I just hope the people involved can move past it.


BREAKING NEWS, From: https://lore.kernel.org/linux-bcache....home.lan/T/#u

From: Kent Overstreet <[email protected]>
To: Stephen Rothwell <[email protected]>
Cc: [email protected], [email protected],
[email protected]
Subject: bcachefs tree for next


Hi Stephen,

Please include a new tree in linux-next:

http://evilpiepirate.org/git/bcachefs.git for-next

I don't see any merge conflicts with the linux-next master branch

Cheers,
Kent
zz0.h7uwiqv5m08zz

[MorrisS.]

That was sarcasm, yes, installing more RAM and disabling SWAP altogether (I've been running swapless for over 20 years now) will fix the issue. Running without SWAP in absolute most cases is more preferrable than with SWAP. If you read anything which contradicts with this statement on the net, it's just plain wrong. SWAP is only necessary when you don't have enough physical RAM to accommodate all the running applications and if that's the case, latency will be horrible. Yeah, Linux uses SWAP for hibernation but it's a brain damaged concept and a bad idea which should have never been implemented this way. Windows has a separate hibernate.sys file for that (or something like that - I don't remember its name).

ok disabling swap is the solution. So why linux operating system has it enabled by default? Soon I'll switch to a SSD.

[Old Grouch]

That was sarcasm, yes, installing more RAM and disabling SWAP altogether (I've been running swapless for over 20 years now) will fix the issue. Running without SWAP in absolute most cases is more preferrable than with SWAP. If you read anything which contradicts with this statement on the net, it's just plain wrong. SWAP is only necessary when you don't have enough physical RAM to accommodate all the running applications and if that's the case, latency will be horrible. Yeah, Linux uses SWAP for hibernation but it's a brain damaged concept and a bad idea which should have never been implemented this way. Windows has a separate hibernate.sys file for that (or something like that - I don't remember its name).

Unfortunately, some low-end systems cannot have their RAM increased to the necessary amount. I'm currently running with a linux swap file three times larger than physical memory, and it fills up (thank-you Firefox). I'd love to be able to add some more RAM. I can't. Finding another system in this form factor is well nigh impossible, or just way, way too expensive.

Coming from my DEC VAX/VMS background, I still think the Linux swap file is a misnomer. AFAIK it's actually a pagefile, as linux doesn't swap out entire process address spaces in one go.

As for hibernating to the swapfile: that is probably an historic artefact from when random-access block devices were smaller, so allocating a significant chunk of a device just for hibernating wasn't a reasonable strategy, and you would likely have a swapfile hanging around that could be used.

[timofonic]

It's extremely easy to call people passive aggressive, and impugn their motives, but I just don't see it having an outsized effect in this circumstance. Certainly I don't think we need a revolt like the CoC thing against Linus a few years back, (and the CoC was not useful at all here, and it was only mentioned during these events briefly and dismissed).

The linux kernel project is extremely mature and extremely unique. That's not to say it's perfect or there's not simple things that they should and can do, but there's limitations on what can be done. Linus has said in the past the roadmap of Linux is what people want to work on and submit. That's pretty much it.

I don't think putting more burden on maintainers and lieutenants to handhold anyone trying to submit code is without cost. Linus has said before they are in that role because they have the quality of having "good taste". [bThat doesn't mean they are optimized for interpersonal skills.[/b]

Both Kent and this forum made quite a big deal of these recent events, so we lack perspective. Time will tell if something will change or come of this.

As for the Linux project, there's really just Linus. There's quite a bit of project support at LF, but not really for managing kernel development directly.

The people who tend to get involved in kernel community stuff include James Bottemley and Ted Ts'o (who got involved but their efforts were fruitless). It's really a massive space, where these senior people tend to not have fixed responsibilities or roles for long time. Morton was like junior Linus way back when but these days he's really just MM stuff. Rik van Riel created kernel newbies but these days he keeps his head down as well.

What you describe aren't signs of an extremely mature project in my humble opinion, quite the contrary.

There should be full-time "handholders" too, in my humble opinion. This could motivate Linux kernel development. From making massive GSoC-like initiatives to more support to innovative developers. Outreachy isn't enough. Linux Foundation is obsessed with cars and printers in GSoC, as you can see here.

FOSS must be not only technical and corporate friendly, but social too.

There needs to be a superviser managers with interpersonal skills. Not to replace current ones, but to support the existing ones.

I consider Linus is somewhat overwhelmed. There's need to be people with fixed responsibilities in the management positions, current situation seems like a mess to me.

There's need for Linux Foundation to support more Linux kernel development instead stupid investments outside Linux kernel, such as a 3D engine and such. Most Linux Foundation efforts seems like joke to me, wasting zillions of money on stupid crap. I think they behave at nearly the same ridicule level of Mozilla Foundation.

I think this is all going to blow over. I just hope the people involved can move past it.

I hope so. But I see it as a sign of management issues too.

BREAKING NEWS, From: https://lore.kernel.org/linux-bcache....home.lan/T/#u

This is very good! I hope things move further and Michael follows it! Thanks for the link!

I really hope it gets merged in linux-next very soon, really...