Btrfs Brings "Pretty Beefy" Changes In Linux 3.2

[haplo602]

Rav,

You're really not doing any service to R4 by trash talking other filesystems and making unverifyable claims about them. It just makes you seem defensive. The kernel devs gave you guys a clear list of the salient problems with Resier4. I would suggest that you guys get to work on it. Trash talking the (time tested and stable) competition isn't going to fix the documented flaws in Reiser4. When the kernel devs deem that the R4 code is good enough to merge, we can revisit the matter.

that is only what's referd to as anecdotal evidence.

take my case. I have a gentoo 64bit system with / being btrfs. 2.6.38 kernel and btrfs v0.19. bought new disk, migrating to GPT+grub2 because I want to boot from lvm.

did some du counts on the different directories, created lvm layout with some overhead, mkbtrfs on all lvols. tar/gz the directories now untaring onto the new filesystem tree I get no space left on /usr ...

df reports:

size 4.0G, used 3.0G, left 4.6m ???? where the hell are 900M if it reports a full fs ? btrfsck, nothing changed. looking for tuning options etc, man pages do not exist or are very short. looking around btrfs wiki, documentation is only marginaly better than man pages on my system.

I recreated all lvols with xfs, untarred the SAME ARCHIVES. now /usr reports 4G size, 4G used, 87M free. but everything untarred into those 4G that did not fit in with btrfs.

never had this problem with reiser or any other fs. what du says in size fits with +/- 1 to 5% of difference into the same type of filesystem on another device. only btrfs does not. WTH ???

/rant

btrfs is nice with all the bells and whistles, but not ready for production systems by a long shot. and it's not only missing btrfsck that's the problem.

EDIT: fixed some typos

[russofris]

I have always wondered what prevents Hans from working on R4 while serving his time. We all agree that he did a truly horrible thing, and that he should be detained, rehabilitated, and pay restitution. I doubt that you will find anyone willing to state the contrary. I just do not see a correlation between his crime and the body of work which he has created. Inmates in prison author books while serving time, why is that any different than authoring code? I understand that his access to the internet is restricted, but does this prevent him from possessing a laptop with a couple VMs and an IDE.

Perhaps he has simply lost motivation due to his circumstances. Perhaps there is another obstacle preventing his involvement (no laptops in prison?).

Does anyone know the deal?

[sbergman27]

I have always wondered what prevents Hans from working on R4 while serving his time.

Computer use, and particularly Internet use, is forbidden in prisons for obvious security reasons. He may be coding with pencil and paper. But that's a pretty severe impediment. Even if he were allowed to mail hand-written patches out for testing, the turn-around time would be days. And besides, R4 wasn't really written by Hans. It was written by Hans' employees. Chris Mason is now at Oracle. Edward is now at RedHat, etc.

I do think that Hans should be allowed to exercise his talents to benefit humanity, if possible. But Hans is exactly the sort of intelligent "game the system" fellow that prison authorities rightly are concerned about.

[sbergman27]

that is only what's referd to as anecdotal evidence.

What is it, exactly that you are calling anecdotal evidence? R4 got a mini code review and came out badly. But any R4 devs which might remain have a clear, if partial, list of unacceptable problems preventing the merging of R4.

btrfs is nice with all the bells and whistles, but not ready for production systems by a long shot. and it's not only missing btrfsck that's the problem.

So far as I know, the only distro which deems btrfs as being production ready is the train wreck known as Fedora.

I have no problem with anyone noting that btrfs still needs some work. Chris Mason would agree. But you won't find Chris trolling forums taking potshots at other filesystems. Insinuations that EXTx will eat your data are particularly ludicrous. The installed base is so great that even 1/10th of one percent of users having had their files eaten by it would swamp the forums.

Also, are you speaking of reiserfs? Or Reiser4? They are completely different filesystems. Hans and Namesys dropped reiserfs as soon as it got merged, forcing Suse to pick up the ball. Reiserfs was basically a data-sieve until Suse threw a huge amount of effort into fixing it. Thus reiserfs (3) is quite stable today.

[haplo602]

What is it, exactly that you are calling anecdotal evidence? R4 got a mini code review and came out badly. But any R4 devs which might remain have a clear, if partial, list of unacceptable problems preventing the merging of R4.



So far as I know, the only distro which deems btrfs as being production ready is the train wreck known as Fedora.

I have no problem with anyone noting that btrfs still needs some work. Chris Mason would agree. But you won't find Chris trolling forums taking potshots at other filesystems. Insinuations that EXTx will eat your data are particularly ludicrous. The installed base is so great that even 1/10th of one percent of users having had their files eaten by it would swamp the forums.

Also, are you speaking of reiserfs? Or Reiser4? They are completely different filesystems. Hans and Namesys dropped reiserfs as soon as it got merged, forcing Suse to pick up the ball. Reiserfs was basically a data-sieve until Suse threw a huge amount of effort into fixing it. Thus reiserfs (3) is quite stable today.

I was refering to my rant about btrfs as anecdotal evidence.

Also I have seen some info about btrfs from Oracle that they want to push it as the main filesystem in their flavor of linux.

[sbergman27]

Does anyone know the deal?

Oh, and BTW, have you seen the latest from Hans? It's a 117 page hand-written motion, in the form of a rant, demanding a new trial. In it, he accuses the entire state of California of being a "hive mind". And asserts that it was a state-wide conspiracy against him that got him convicted. He reprises his claim that his former attorney, William DuBois, has excessive levels of the hormone oxytocin, which he believes causes people to enjoy betraying others. And wants a blood test done.

(And oh yeah. He's also suing the state of California for 3 trillion dollars. A trillion for him. And a trillion each for his 2 children. Meanwhile, his children and their guardians are suing Hans for whatever they can get since they are living on a very low income in Russia.)

But the core of his "argument", to the extent that it can be called that, is that DuBois forced him to take the stand and hang himself. Anyone who was following the news, back in 2008, knows very well that back then he was complaining that DuBois *wouldn't* let him take the stand, but Hans insisted. Had Hans not done that, he would most likely have gotten off scott-free.

At this point, I really wouldn't object to them transferring Hans to a mental hospital. I've never liked him. But I'm actually starting to feel kinda sorry for the guy.

[sbergman27]

I was refering to my rant about btrfs as anecdotal evidence.

Oh. Well, that's cool. Please forgive me for being in "aggressive mode" today. ;-)

Also I have seen some info about btrfs from Oracle that they want to push it as the main filesystem in their flavor of linux.

I think pretty much everyone does. When it's ready. But to be honest, I'm not champing at the bit. EXT4, with delayed allocation turned off, is a damned good filesystem. The main looming problem, as I see it, is its 16TiB size limit. But that's not so much an EXT4 thing as a user space utilities thing. EXT4 is good for a full 48 bits. It's only ext2utils that peters out at 16 TiB.

[liam]

Can you explain this a bit more? I thought BTRFS detected and repaired silent corruption? But you say no?

It should if you include the scrub patch and have a duplicate to base the repair on (since it is COW you MAY not need a mirror to recover corrupted data, but a mirror should make it MUCH more likely).
Does anyone know if the scrub patch was accepted? It's pretty old now, but I haven't heard anything about it...

[liam]

What is it, exactly that you are calling anecdotal evidence? R4 got a mini code review and came out badly. But any R4 devs which might remain have a clear, if partial, list of unacceptable problems preventing the merging of R4.



So far as I know, the only distro which deems btrfs as being production ready is the train wreck known as Fedora.

I have no problem with anyone noting that btrfs still needs some work. Chris Mason would agree. But you won't find Chris trolling forums taking potshots at other filesystems. Insinuations that EXTx will eat your data are particularly ludicrous. The installed base is so great that even 1/10th of one percent of users having had their files eaten by it would swamp the forums.

Also, are you speaking of reiserfs? Or Reiser4? They are completely different filesystems. Hans and Namesys dropped reiserfs as soon as it got merged, forcing Suse to pick up the ball. Reiserfs was basically a data-sieve until Suse threw a huge amount of effort into fixing it. Thus reiserfs (3) is quite stable today.

Umm, why are you calling fedora a train wreck? They have said no btrfs (by default) until a repair tool exists (so, IOW, it is NOT ready for production). The users of fedora understand that it is really for testing new software but, amazingly, that doesn't mean it isn't horribly unstable. A, great, QA team exists, and makes sure each release is as stable as possible. That is why they don't keep to hard release schedules. Having du not work doesn't mean btrfs is broken, necessarily, but it MAY mean that du needs to be updated to work with btrfs.

[haplo602]

Oh. Well, that's cool. Please forgive me for being in "aggressive mode" today. ;-)



I think pretty much everyone does. When it's ready. But to be honest, I'm not champing at the bit. EXT4, with delayed allocation turned off, is a damned good filesystem. The main looming problem, as I see it, is its 16TiB size limit. But that's not so much an EXT4 thing as a user space utilities thing. EXT4 is good for a full 48 bits. It's only ext2utils that peters out at 16 TiB.

No problem I know my ENglish is not the best and I do have problems getting my points across in a clear way.

As to the FS limit, my occupation is UX system admin for Oracle/SAP running systems. We usualy split filesystems at the 2 or 4TB limit to minimise damage in case of corruptions (we DO have several 10s of TB large databases). I don't see a practical use for a 16TB fs. It looks nice on marketing slides, but practical use is none.