[opensuse-factory] Proposal: Change in default fs for releases >= 10.2
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Hi all - We've been using ReiserFS as our default installation file system for the last 6-7 years now, and it's served us well in that time. Unfortunately, there are a number of problems with it, some purely technical, some more related to maintenance. I'll outline a few of the larger issues and offer my solution as a conclusion. ReiserFS has serious scalability problems. David Chinner's talk at OLS really underscored the problem well for a single, large, high bandwidth file system. While I realize that XFS-style scalability isn't a real goal for most users, ans isn't a target workload for reiserfs, the scalability problems are real. ReiserFS uses the BKL for synchronization everywhere, and since it's system-global lock, the problem doesn't go away when you split the file system into smaller ones. Lock contention alone is one problem, but it's made worse by cache bouncing between processors on larger systems. ReiserFS has serious performance problems with extended attributes and ACLs. (Yes, this one is my own fault, see numerous flamewars on lkml and reiserfs-list for my opinion on this.) xattrs are backed by normal files rooted in a hidden directory structure. This is bad for performance and in rare cases is deadlock prone due to lock inversions between pdflush and the xattr code. The quota code gets around this, but the fix would result in huge amounts of wasted space with ReiserFS. With increasing deployment of SLES as samba servers, and perhaps NFSv4 servers, the use of extended attributes will only increase. ReiserFS has a small and shrinking development community. Right now, the only developers really working with ReiserFS are Chris Mason, Jan Kara (internally), a rotating member of Hans Reiser's team, and myself. All of us have projects we're very much more interested in than working with ReiserFS. While Jan and I will be continuing to support ReiserFS for SUSE, Hans is increasingly (hard to believe) pushing people to use reiser4. Chris has moved on to Oracle and has expressed his opinions on leaving ReiserFS behind. ReiserFS v3 is a dead end. Hans has been pushing reiser4 for years now and declared Reiser3 in maintenance mode. Any changes that aren't bug fixes are met with violent resistance. Reiser4 is not an incremental update and requires a reformat, which is unreasonable for most people. Reiser3 lacks a number of features that other file systems either have or are adding soon, such as extents and growth beyond current limits. Since it's in maintenance mode, that's unlikely to change. I view reiser4 as an interesting research file system, but that's about as far as it goes. I've been unimpressed with its stability so far. I don't know how advanced the recovery tools are yet, but I suspect that the complexity of the format and the ability to essentially define the format on-the-fly with plugins will make a useful fsck extremely difficult. The solution for replacing an aging file system isn't to switch to a brand new unproven file system, but rather a proven one with a clear upgrade path. That file system is ext3. Ext3's performance in some situations may not be on par with Reiser3, but it scales better and Andi mentioned the other day that there is quite a bit of research going into improving the locking and general performance of ext3 going on right now, and since reiser3 is stagnant, I don't doubt they'll pass them soon. Ext3 has a much larger development community out there. Most other distributions use ext3 as their default file system, so bugs that don't end up getting reported to us and are fixed by other developers, we get for free - most Reiser3 fixes originate from Chris or I. Ext3 has a clear upgrade path. There is quite a bit of interest in the community in improving ext3, and ext4 is already under development. Like the upgrade path from ext2 to ext3, the path to ext4 is clearly defined. Existing file systems can be updated easily, and new files will be able to take advantage of the new features. Features already written and queued up include extents, a 64-bit journal, and 64-bit file sizes. Most of the institutional knowledge of reiserfs is bouncing around in my head. Jan has been getting his hands dirty a little bit, but beyond that, finding additional developers with reiserfs experience will be extremely difficult and I'd call training additional developers a wasted effort. Since reiserfs is in maintenance mode, the effort needed to continue to support it in future releases should be shrinking. To be clear, my long term goal is to use OCFS2 (or another CFS if one shows a clear adoption advantage) for the root file system. This would enable single-instance clustering at both the physical and the virtual distribution level and get us ease of management and flexibility in HA deployments. Realistically, though, desktop users are likely to continue to use ext[34] for the foreseeable future. Until we have OCFS2 (and the rest of the distribution) ready for such a deployment described above on larger servers, ext3 would be a suitable choice across the board. - -Jeff - -- Jeff Mahoney SUSE Labs -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.2 (GNU/Linux) Comment: Using GnuPG with SUSE - http://enigmail.mozdev.org iD8DBQFFCSu+LPWxlyuTD7IRAgJoAJ49w+Ml6/cf4HqLRoHAb7L7S3lDMQCfafIL JkIApJTPoS6sytFVaDI++KE= =mFFG -----END PGP SIGNATURE----- --------------------------------------------------------------------- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org For additional commands, e-mail: opensuse-factory+help@opensuse.org
Thanks for the detailed explanations and discussion. Out of interest, is XFS also under consideration? I don't have experience of either ext3 nor XFS on my personal or work machines, but I've heard many good things about XFS from friends, and from reading comparisons. I'd be interested in knowing what your current feelings are regarding the relative strengths and weaknesses of the two. --------------------------------------------------------------------- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org For additional commands, e-mail: opensuse-factory+help@opensuse.org
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 William Gallafent wrote:
Thanks for the detailed explanations and discussion.
Out of interest, is XFS also under consideration? I don't have experience of either ext3 nor XFS on my personal or work machines, but I've heard many good things about XFS from friends, and from reading comparisons.
I'd be interested in knowing what your current feelings are regarding the relative strengths and weaknesses of the two.
Technically, XFS is a good file system, but my primary concern with it is one of support. XFS is far more complex than ext3, and it weighs in at over 5 times the lines of code that ext3 does. Beyond that, there is far more community adoption and support behind ext3. Inside SUSE, we just don't have the engineering experience with it and that means more time away from fixing other bugs and implementing new features. This proposal is only to change the default file system. Ext3 does what most users need well. It's well tested, is fairly easy to understand, and is extremely well supported by the community. Users who have workloads that require, or are even just curious about, other file systems, are always welcome to use them. I'm not looking to, or at all interested in, removing file systems from the list of possibilities. ReiserFS will still be available, as will XFS. - -Jeff - -- Jeff Mahoney SUSE Labs -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.2 (GNU/Linux) Comment: Using GnuPG with SUSE - http://enigmail.mozdev.org iD8DBQFFCXaELPWxlyuTD7IRAlTQAJ4kcTR5AwgEzHJqbOvSPG6ZcZjpqgCcC/ZO eL5ZncH/1TpL8H0b7m3os7M= =ZsJ3 -----END PGP SIGNATURE----- --------------------------------------------------------------------- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org For additional commands, e-mail: opensuse-factory+help@opensuse.org
Jeff Mahoney a écrit :
larger servers, ext3 would be a suitable choice across the board.
very clever post, and you seems the better placed to have an opinion :-(). I use reiserfs from the beginning, with little problems and I have tried ext3 with some problems, mostly when switching from ext2 to ext3 which is not as riskless as it's said. but this is not of great importance. I just quote this to say that it must be said that ext2 and ext3 are _not_ the same file system. For example I have a native windows XP ext2 driver that don't read ext3... jdd -- http://www.dodin.net http://dodin.org/galerie_photo_web/expo/index.html http://lucien.dodin.net http://fr.susewiki.org/index.php?title=Gérer_ses_photos --------------------------------------------------------------------- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org For additional commands, e-mail: opensuse-factory+help@opensuse.org
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 jdd wrote:
Jeff Mahoney a écrit :
larger servers, ext3 would be a suitable choice across the board.
very clever post, and you seems the better placed to have an opinion :-().
I use reiserfs from the beginning, with little problems and I have tried ext3 with some problems, mostly when switching from ext2 to ext3 which is not as riskless as it's said. but this is not of great importance.
I just quote this to say that it must be said that ext2 and ext3 are _not_ the same file system. For example I have a native windows XP ext2 driver that don't read ext3...
When ext3 split from ext2, they essentially were the same file system. The magic number would have been different, but the file systems were otherwise the same. Since then, ext2 has been placed in maintenance mode and additional features have been added to ext3, some of which are incompatible. I view this as a *good thing*. Ext2 may not be able to read ext3 file systems, but ext3 can certainly read ext2 file systems. This is intended as an upgrade path, not to be be backward compatible with the old code. It is a superset of ext2. The old file system code remains stable for those who still want to use it, while the new file system code gets the new features. This is how it will also be with ext4. I'm not sure what problems you had when upgrading from ext2, but tune2fs - -j should have only changed the magic number and added a journal. - -Jeff - -- Jeff Mahoney SUSE Labs -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.2 (GNU/Linux) Comment: Using GnuPG with SUSE - http://enigmail.mozdev.org iD8DBQFFCW+nLPWxlyuTD7IRAi2CAJ9VcZRilCFntVs7B+bJGGCoPdeNhwCfdMuC oyIGzXYm4qIGgbucFpG3WJQ= =1aWY -----END PGP SIGNATURE----- --------------------------------------------------------------------- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org For additional commands, e-mail: opensuse-factory+help@opensuse.org
Jeff Mahoney a écrit :
I'm not sure what problems you had when upgrading from ext2, but tune2fs - -j should have only changed the magic number and added a journal.
I use an usb drive (300Gb) as swap between windows and Linux. I use an etx2 native driver for XP that allows ext2 partition to be read exactly as NTFS ones, very well done and really usefull; At first I formatted my drive with ext2, it worked, then I tried ext3 (ext2 with 300Gb, don't do a fsck too often :-!) but this one was not read by the win driver, so I switched back to ext2... and loose the data. these data where garbage (test bed), but this make me feel it's better not rely on the ext2<>ext3 compatibility (what is not a real problem, I live with ext2) jdd -- http://www.dodin.net http://dodin.org/galerie_photo_web/expo/index.html http://lucien.dodin.net http://fr.susewiki.org/index.php?title=Gérer_ses_photos --------------------------------------------------------------------- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org For additional commands, e-mail: opensuse-factory+help@opensuse.org
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 jdd wrote:
Jeff Mahoney a écrit :
I'm not sure what problems you had when upgrading from ext2, but tune2fs - -j should have only changed the magic number and added a journal.
I use an usb drive (300Gb) as swap between windows and Linux. I use an etx2 native driver for XP that allows ext2 partition to be read exactly as NTFS ones, very well done and really usefull;
At first I formatted my drive with ext2, it worked, then I tried ext3 (ext2 with 300Gb, don't do a fsck too often :-!) but this one was not read by the win driver, so I switched back to ext2... and loose the data.
these data where garbage (test bed), but this make me feel it's better not rely on the ext2<>ext3 compatibility (what is not a real problem, I live with ext2)
There is not ext2<>ext3 compatibility. The compatibility is one-way, but can work the other way until you enable an incompatible feature. If you have an ext3 implementation, you'll be able to read ext2 and ext3. If you have an ext4 implementation (when it's complete), you'll be able to read ext2 and ext3 file systems. Ext3 won't be able to read all ext4 file systems, just as ext2 can't always read ext3 file systems. Each successive version is a superset of the earlier version. It's an upgrade path. Once you add incompatible features, like hashed directories, for example, you can't mount it on an older implementation. This is known, and why these features need to be explicitly enabled. That said, I have no idea what the windows ext2 driver is doing, and I don't consider it not understanding ext3 file systems a show stopper. - -Jeff - -- Jeff Mahoney SUSE Labs -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.2 (GNU/Linux) Comment: Using GnuPG with SUSE - http://enigmail.mozdev.org iD8DBQFFCnRqLPWxlyuTD7IRAjjjAJ46ZtatUcXd+PtGt6+3jlbqtxtKFQCgmdjb T8I5kI6PE96pJhb4pGky0D4= =l4PR -----END PGP SIGNATURE----- --------------------------------------------------------------------- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org For additional commands, e-mail: opensuse-factory+help@opensuse.org
Jeff Mahoney a écrit :
don't consider it not understanding ext3 file systems a show stopper.
it's not a stopper at all :-) however, having a Windows XP driver for a Linux file system is really a plus in our days of mobile hard drives. jdd -- http://www.dodin.net http://dodin.org/galerie_photo_web/expo/index.html http://lucien.dodin.net http://fr.susewiki.org/index.php?title=Gérer_ses_photos --------------------------------------------------------------------- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org For additional commands, e-mail: opensuse-factory+help@opensuse.org
Am Freitag, 15. September 2006 11:37 schrieb Jeff Mahoney:
That said, I have no idea what the windows ext2 driver is doing, and I don't consider it not understanding ext3 file systems a show stopper.
I use http://www.fs-driver.org/ at home and can read ext3 file system just fine (if they were cleanly unmounted). Greetings, Stephan --------------------------------------------------------------------- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org For additional commands, e-mail: opensuse-factory+help@opensuse.org
Am Donnerstag, 14. September 2006 12:15 schrieb Jeff Mahoney:
Hi all - [...] The solution for replacing an aging file system isn't to switch to a brand new unproven file system, but rather a proven one with a clear upgrade path. That file system is ext3.
Nice to hear that. I'm also a ReiserFS user from the first minute (Suse 6.4) and I never had problems with it. I also belive, that with a little support from the community reiser4 would have a better future, but that's politics. As far as the change from ReiserFS to Ext3 is concerned however, I agree to do it. I used to do some stupid fs benchmarks with copying kernel source from one partition to another and with SL 10.0 things changed. ReiserFS always used to be faster than Ext3 for small files. And as most Linux files are small <= 100 k, Reiser did the work better. But from 10.0 on, Ext3 has gone faster (ReiserFS has gone slower) and it even performs now 10-20% better, than ReiserFS. These results are from yesterday: Copying 260 MByte from Partiton A to Partition B Ext3 ReiserFS XFS SL 10.1 53s 65s 210s Dapper 54s 71s 200s Probably this speed down from ReiserFS is also due to the barrier flush option, at least I remember, that on SLES9 it made ReiserFS very slow. As you can see, XFS is not an alternative for the root filesystem nighter, as it handles small files very slow. -- Mit freundlichen Grüßen, Marcel Hilzinger Linux New Media AG Süskindstr. 4 D-81929 München Tel: +49 (89) 99 34 11 0 Fax: +49 (89) 99 34 11 99 --------------------------------------------------------------------- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org For additional commands, e-mail: opensuse-factory+help@opensuse.org
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Marcel Hilzinger wrote:
Am Donnerstag, 14. September 2006 12:15 schrieb Jeff Mahoney:
Hi all - [...] The solution for replacing an aging file system isn't to switch to a brand new unproven file system, but rather a proven one with a clear upgrade path. That file system is ext3.
Nice to hear that. I'm also a ReiserFS user from the first minute (Suse 6.4) and I never had problems with it. I also believe, that with a little support from the community reiser4 would have a better future, but that's politics.
As far as the change from ReiserFS to Ext3 is concerned however, I agree to do it. I used to do some stupid fs benchmarks with copying kernel source from one partition to another and with SL 10.0 things changed. ReiserFS always used to be faster than Ext3 for small files. And as most Linux files are small <= 100 k, Reiser did the work better. But from 10.0 on, Ext3 has gone faster (ReiserFS has gone slower) and it even performs now 10-20% better, than ReiserFS. These results are from yesterday:
Copying 260 MByte from Partiton A to Partition B
Ext3 ReiserFS XFS SL 10.1 53s 65s 210s Dapper 54s 71s 200s
Interesting numbers, but it really depends a lot on the tests you make.
Things tend to change a lot when under heavy load, when you have
concurrent access, when the drive is >80% full or not, etc...
I've rather recently (just a few months ago) seen some seriously done
benchmark where reiserfs still had the edge, by a significant margin.
Not sure I'll be able to find it again, take my word for it ;)
XFS scales best in terms of parallelization (number of CPUs/cores) but
is a very bad choice for a typical desktop, also because it is said to
use more CPU time than the other filesystems.
- From what I've read, XFS is prolly the best option e.g. for streaming
large files when you have a lot of CPUs, but it scales down very badly.
ext3 is said to have its performance go down the gutter once your disk
is 70/80% full, I wonder if that's still accurate with recent kernels.
AFAIK (might be wrong though, haven't use ext3 since ages), ext3 still
isn't capable of being resized while mounted, something reiserfs is able
to do since quite some time (only growing though) - a *very* useful
feature, especially in combination with LVM.
Right/wrong ?
[...]
- --
-o) Pascal Bleser http://linux01.gwdg.de/~pbleser/
/\\
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Pascal Bleser wrote:
ext3 is said to have its performance go down the gutter once your disk is 70/80% full, I wonder if that's still accurate with recent kernels.
I'd like to see benchmarks for this.
AFAIK (might be wrong though, haven't use ext3 since ages), ext3 still isn't capable of being resized while mounted, something reiserfs is able to do since quite some time (only growing though) - a *very* useful feature, especially in combination with LVM. Right/wrong ?
It should be capable of performing online resize operations. If not in the current kernel, I've seen patches that implement it. - -Jeff - -- Jeff Mahoney SUSE Labs -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.2 (GNU/Linux) Comment: Using GnuPG with SUSE - http://enigmail.mozdev.org iD8DBQFFCeo4LPWxlyuTD7IRAqpgAJ43VbNLWH0Uqucmx27pkaPf2phq8wCgkblU tDmdz74AQHksdIbDFjz7fP8= =WUgY -----END PGP SIGNATURE----- --------------------------------------------------------------------- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org For additional commands, e-mail: opensuse-factory+help@opensuse.org
AFAIK (might be wrong though, haven't use ext3 since ages), ext3 still isn't capable of being resized while mounted, something reiserfs is able to do since quite some time (only growing though) - a *very* useful feature, especially in combination with LVM. Right/wrong ?
It should be capable of performing online resize operations. If not in the current kernel, I've seen patches that implement it.
There's ext2online in e2fsprogs-package - but it lists a lot of caveats. I tested it briefly and must say that xfs_grow and resize_reiserfs look more 'serious' -- but that also needs more testing... Regards Heiko --------------------------------------------------------------------- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org For additional commands, e-mail: opensuse-factory+help@opensuse.org
Op donderdag 14 september 2006 12:15, schreef Jeff Mahoney:
Hi all - To be clear, my long term goal is to use OCFS2 (or another CFS if one shows a clear adoption advantage) for the root file system. This would enable single-instance clustering at both the physical and the virtual distribution level and get us ease of management and flexibility in HA deployments. Realistically, though, desktop users are likely to continue to use ext[34] for the foreseeable future. Until we have OCFS2 (and the rest of the distribution) ready for such a deployment described above on larger servers, ext3 would be a suitable choice across the board.
-Jeff
Just some support cause I really don't know anything about filesystems. It seems reasonable to switch. Azerion --------------------------------------------------------------------- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org For additional commands, e-mail: opensuse-factory+help@opensuse.org
Hello, Jeff Mahoney wrote:
Ext3 has a much larger development community out there. Most other distributions use ext3 as their default file system, so bugs that don't end up getting reported to us and are fixed by other developers, we get for free - most Reiser3 fixes originate from Chris or I.
Changing to ext3 would be a very welcome change for Pegasos PPC users. The OpenFirmware implementation of these machines can read files and boot from ext2 / ext3, but not reiserfs, just for the above reasons. Making ext3 default would make our lives a lot more easy and required documentation a lot shorter :-) Bye, -- CzP http://peter.czanik.hu/ --------------------------------------------------------------------- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org For additional commands, e-mail: opensuse-factory+help@opensuse.org
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Peter Czanik wrote:
Hello,
Jeff Mahoney wrote:
Ext3 has a much larger development community out there. Most other distributions use ext3 as their default file system, so bugs that don't end up getting reported to us and are fixed by other developers, we get for free - most Reiser3 fixes originate from Chris or I.
Changing to ext3 would be a very welcome change for Pegasos PPC users. The OpenFirmware implementation of these machines can read files and boot from ext2 / ext3, but not reiserfs, just for the above reasons. Making ext3 default would make our lives a lot more easy and required documentation a lot shorter :-) Bye,
Are these old world machines? I used to develop on a powermac and ended up developing the reiserfs support in yaboot. - -Jeff - -- Jeff Mahoney SUSE Labs -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.2 (GNU/Linux) Comment: Using GnuPG with SUSE - http://enigmail.mozdev.org iD8DBQFFCeroLPWxlyuTD7IRAhpZAJ9POltMIlsiKsYuF9cw6Uh9e+KJFQCgiSyw RiUM10WjnFKXecRyYVlCQSc= =GJ/z -----END PGP SIGNATURE----- --------------------------------------------------------------------- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org For additional commands, e-mail: opensuse-factory+help@opensuse.org
Hello, Jeff Mahoney wrote:
Changing to ext3 would be a very welcome change for Pegasos PPC users. The OpenFirmware implementation of these machines can read files and boot from ext2 / ext3, but not reiserfs, just for the above reasons. Making ext3 default would make our lives a lot more easy and required documentation a lot shorter :-) Bye, Are these old world machines? I used to develop on a powermac and ended up developing the reiserfs support in yaboot.
No, these are not Macs, but a completely separate type of PPC machines, designed by bPlan from ground up: http://www.pegasosppc.com/pegasos.php -- CzP http://peter.czanik.hu/ --------------------------------------------------------------------- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org For additional commands, e-mail: opensuse-factory+help@opensuse.org
Am Thursday 14 September 2006 23:16 schrieb Peter Czanik:
Hello,
Jeff Mahoney wrote:
Ext3 has a much larger development community out there. Most other distributions use ext3 as their default file system, so bugs that don't end up getting reported to us and are fixed by other developers, we get for free - most Reiser3 fixes originate from Chris or I.
Changing to ext3 would be a very welcome change for Pegasos PPC users. The OpenFirmware implementation of these machines can read files and boot from ext2 / ext3, but not reiserfs, just for the above reasons. Making ext3 default would make our lives a lot more easy and required documentation a lot shorter :-) Bye,
It would be enough to change the default on PPC systems only in this case. Olaf, does this make sense for PPC ? bye adrian -- Adrian Schroeter SUSE Linux Products GmbH, Maxfeldstr. 5, 90409 Nuernberg, Germany email: adrian@suse.de --------------------------------------------------------------------- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org For additional commands, e-mail: opensuse-factory+help@opensuse.org
participants (10)
-
Adrian Schröter -
Azerion -
Heiko Helmle -
jdd -
Jeff Mahoney -
Marcel Hilzinger -
Pascal Bleser -
Peter Czanik -
Stephan Kulow -
William Gallafent