I am going to install 42.3 on a computer whose hard disk I want to erase completely to change partitioning. This gives me the chance to choose the file system. So far, I used ext4 only. But recently I read more about XFS and btrfs. But I know too little about them. Are they better than ext4? Or worse? Can you tell me something about their advantages and possible defects? Are there bugs? Best regards, Wolfgang -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
Op maandag 26 augustus 2019 18:09:51 CEST schreef Wolfgang Mueller:
I am going to install 42.3 Don't. It's out of maintenance. Install 15.0 or even better 15.1 on a computer whose hard disk I want to erase completely to change partitioning. This gives me the chance to choose the file system.
So far, I used ext4 only. But recently I read more about XFS and btrfs. But I know too little about them. Are they better than ext4? Or worse? Can you tell me something about their advantages and possible defects? Are there bugs?
Best regards, Wolfgang
-- Gertjan Lettink a.k.a. Knurpht openSUSE Board Member openSUSE Forums Team -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
Op maandag 26 augustus 2019 18:09:51 CEST schreef Wolfgang Mueller:
I am going to install 42.3 on a computer whose hard disk I want to erase completely to change partitioning. This gives me the chance to choose the file system.
So far, I used ext4 only. But recently I read more about XFS and btrfs. But I know too little about them. Are they better than ext4? Or worse? Can you tell me something about their advantages and possible defects? Are there bugs?
Best regards, Wolfgang Clicked Send too soon :)
IME btrfs is rockstable. Combined with snapper snapshots it allows you to roll back to a previous state of the fs, f.e. when updates mess up things. Shouldn't happen, but still. I use the feature a lot: install something to support a user, when done rollback to before the install of that something. Awesome. -- Gertjan Lettink a.k.a. Knurpht openSUSE Board Member openSUSE Forums Team -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 26/08/2019 12:23, Knurpht-openSUSE wrote:
IME btrfs is rockstable. Combined with snapper snapshots it allows you to roll back to a previous state of the fs, f.e. when updates mess up things. Shouldn't happen, but still. I use the feature a lot: install something to support a user, when done rollback to before the install of that something. Awesome.
Gee WOW! And here an I doing simply "zypper install/zypper remove" to achieve much the same thing. It may be that there are other users of that file system or other changes made to the file system that rolling back the image would corrupt. Of course there's going to be "more to it than that", isn't there? You may have a file-system-per-user multi user system. You may have an arrangement where you can cherry=pick that gets rolled back and what doesn't. But from an administrative POV it still seems like a lot of work. Of course we could end up with a multi-user machine where there is a file-system-per user and the user has full admin control, including doing BtrFS rollbacks to suit his or her own wishes. My ISP, Dreamhost, supports that; it is a virtual machine per user arrangement. Of course when you've got virtual machines to play with you can install 42.3 and earlier to your heart's content to do whatsoever amuses you. personally, I found that on a single user, personal machine having btrFS, 3spcially saving images for every zypper update change, was too much overhead. Perhaps I have better admen discipline; perhaps I have better backup discipline and technology, perhaps I 'snapshot' my edits and have scripts for the RCS when doing 'development'. Yes, I found I could configure BtrFS to deal with the 'exceptions', disable the bits I didn't want it to do. But the end result was pretty much disabling BtrFS. On top of that, I use LVM and BtrFS has a "one file system to rule them all" attitude towards drive and is, in effect, a "competitor" to LVM. LVM lets me use different file systems for different parts of what I want. It lets me compare and take metrics. I use Ext4 (and Ext3 and ext2), ReiserFS, XFS and JFS. As it happens, my preference these days is towards JFS over ReiserFS since Reiser4 is not part of the distribution. I am, however, running a personal system. If I were running a commercial system, what amounts to a supermicro replacement for what used to be the raised-floor IT department of a corporation in an IBM mainframe, then things are going to be different. The way the old IBM mainframes worked was very admin-comprehensive and yes, they had the roll-back of changes mechanism that we have in BtrFS. But a single user, a home system, things are going to be different. Perhaps you are doing a lot of development and find it easier to have BtrFS snapshot every 5 minutes than have software that does that, have a revision control system. Of course a RCS can deal with 'branches' and comments in a way that BtrFS cannot. Needs must drive. Of course you can look at it this way. Ext4 is supposed to be on a development path that will allow 'snapper', the guts of what does the BtrFS snapshhot, to give ext4 the same capability. When it will get here I don't know, but going for Ext4 hedges your bets. -- A: Yes. > Q: Are you sure? >> A: Because it reverses the logical flow of conversation. >>> Q: Why is top posting frowned upon? -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On Mon, 26 Aug 2019 16:03:38 -0400 Anton Aylward <opensuse@antonaylward.com> wrote:
The way the old IBM mainframes worked was very admin-comprehensive and yes, they had the roll-back of changes mechanism that we have in BtrFS.
I generally agree with what you're saying, but I can't help pointing out that GEORGE3 had versioned file storage back in the early seventies. Back when printers kept you toasty in winter. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
Anton Aylward wrote:
On 26/08/2019 12:23, Knurpht-openSUSE wrote:
IME btrfs is rockstable. Combined with snapper snapshots it allows you to roll back to a previous state of the fs, f.e. when updates mess up things. Shouldn't happen, but still. I use the feature a lot: install something to support a user, when done rollback to before the install of that something. Awesome.
Gee WOW! And here an I doing simply "zypper install/zypper remove" to achieve much the same thing.
So zypper allows you to roll back to packages that are no longer in the repository? That was the reason for the only rollback I've done in 3 years TW (broken X packages). One thing I sometimes use is the simple way of checking which files have changed, and to easily do a diff between versions (e.g., of config files).
personally, I found that on a single user, personal machine having btrFS, 3spcially saving images for every zypper update change, was too much overhead.
I sort of agree there, that's why I removed/commented <!-- <solvable match="w">*</solvable> --> in /etc/snapper/zypp-plugin.conf. With that, I find the overhead acceptable, and the benefits win ;^> -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 30/08/2019 05:08, Peter Suetterlin wrote:
Anton Aylward wrote:
On 26/08/2019 12:23, Knurpht-openSUSE wrote:
IME btrfs is rockstable. Combined with snapper snapshots it allows you to roll back to a previous state of the fs, f.e. when updates mess up things. Shouldn't happen, but still. I use the feature a lot: install something to support a user, when done rollback to before the install of that something. Awesome.
Gee WOW! And here an I doing simply "zypper install/zypper remove" to achieve much the same thing.
So zypper allows you to roll back to packages that are no longer in the repository? That was the reason for the only rollback I've done in 3 years TW (broken X packages).
Well MY use case that annoyed me immensely was that snapper took a snapshot EVERY TIME I ran zypper to install a package to try out, and 'addition'; that goes for kernel updates, and it's why I update kernel rather than patch the kernel!. I can't say that the "no longer in the repository" situation is one I've met ... ... EXCEPT ... ... when I've been forced to upgrade, for example from 42.3 to the 15.x because of the repositories going away. No so much "no longer IN the repository" as "no longer THE repository"!
One thing I sometimes use is the simple way of checking which files have changed, and to easily do a diff between versions (e.g., of config files).
Ah. Well perhaps you've never used VAX VMS ... Yes, I've done diff between the config files; zypper has this thing that it won't simply update them if the original has been changed, so I have lots of later .rpm files under /etc that I can 'diff' toe see what updates I'm supposed to be getting BEFORE I install them, often using 'sdiff' to pick and choose between the old and the new. Nothing whatsoever to do with snapper, thank you! In other areas I set up my applications to take backups or snapshots as I'm working; in a development IDE I have a RCS system in place. Its not simply about 'snapshots' its often about documenting the snapshots and sometimes branching as ideas develop and I try them out. Failed ideas are valuable too and need to be preserved. And WHY they failed needs to be documented. A good RCS allows for all this. So sorry, its not just snapper that annoyed the hell out of me, even after much configuration, it was BtrFS as well.
personally, I found that on a single user, personal machine having btrFS, 3spcially saving images for every zypper update change, was too much overhead.
I sort of agree there, that's why I removed/commented <!-- <solvable match="w">*</solvable> --> in /etc/snapper/zypp-plugin.conf.
Right. I configured the hell out of that, removed modules and more, but as I said, the root of the problem was BtrFS. I got rid of that, am using Ext4 for my, mostly, stable RootFS. -- A: Yes. > Q: Are you sure? >> A: Because it reverses the logical flow of conversation. >>> Q: Why is top posting frowned upon? -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
Anton Aylward wrote:
... when I've been forced to upgrade, for example from 42.3 to the 15.x because of the repositories going away. No so much "no longer IN the repository" as "no longer THE repository"!
Just in case you are not aware - some of our mirrors still serve older repositories, e.g. ftp.gwdg.de : ftp://ftp.gwdg.de/pub/opensuse/discontinued/ -- Per Jessen, Zürich (27.9°C) member, openSUSE Heroes. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
Per Jessen wrote:
Anton Aylward wrote:
... when I've been forced to upgrade, for example from 42.3 to the 15.x because of the repositories going away. No so much "no longer IN the repository" as "no longer THE repository"!
Just in case you are not aware - some of our mirrors still serve older repositories, e.g. ftp.gwdg.de :
ftp://ftp.gwdg.de/pub/opensuse/discontinued/
Indeed, I'm using that a lot, too. But I was rather speaking of Tumbleweed. If you do a zypper dup and then find that some package has issues there's a substantial chance that the rpm for the version you had before is already gone... Just do a 'zypper ref' and then a 'zypper search -si'. You'll see quite a lot of them marked as 'System Package' - because there's no more RPM for it in the repo... -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
30.08.2019 18:32, Peter Suetterlin пишет:
Per Jessen wrote:
Anton Aylward wrote:
... when I've been forced to upgrade, for example from 42.3 to the 15.x because of the repositories going away. No so much "no longer IN the repository" as "no longer THE repository"!
Just in case you are not aware - some of our mirrors still serve older repositories, e.g. ftp.gwdg.de :
ftp://ftp.gwdg.de/pub/opensuse/discontinued/
Indeed, I'm using that a lot, too.
But I was rather speaking of Tumbleweed. If you do a zypper dup and then find that some package has issues there's a substantial chance that the rpm for the version you had before is already gone...
http://download.opensuse.org/history/ https://github.com/boombatower/tumbleweed-cli -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
Andrei Borzenkov wrote:
Interesting, I didn't know about those. Will need some reading. Thanks! -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
Anton Aylward wrote:
Well MY use case that annoyed me immensely was that snapper took a snapshot EVERY TIME I ran zypper to install a package to try out, and 'addition'; that goes for kernel updates, and it's why I update kernel rather than patch the kernel!.
The snapshots for every call is cured by removing the wildcard match rule, of course you can as well just remove the whole snapper-zypp-plugin package.
I can't say that the "no longer in the repository" situation is one I've met ...
As mentioned in the other post, that is a somewhat specific issue for Tumbleweed with its permanent updates. Indeed, I don't really remember this for the Leaps, they most of the time have several older versions still in.
Ah. Well perhaps you've never used VAX VMS ...
I did. But the multi-version-filesystem isn't something I'm really missing.
So sorry, its not just snapper that annoyed the hell out of me, even after much configuration, it was BtrFS as well.
Everyone is free to pick his favorites. I didn't want to convince you of anything, just wanted to point out that some issues can easily be solved.. :) -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 30/08/2019 11:45, Peter Suetterlin wrote:
Anton Aylward wrote:
Well MY use case that annoyed me immensely was that snapper took a snapshot EVERY TIME I ran zypper to install a package to try out, and 'addition'; that goes for kernel updates, and it's why I update kernel rather than patch the kernel!.
The snapshots for every call is cured by removing the wildcard match rule, of course you can as well just remove the whole snapper-zypp-plugin package.
Yes, that's what I did. But this was just another point in a long series of annoyances that went with the whole attitude that BtrFS brought along. I got rid of the 'root cause' - getting rid of BtrFS got rid of all my snapper problems.
I can't say that the "no longer in the repository" situation is one I've met ...
As mentioned in the other post, that is a somewhat specific issue for Tumbleweed with its permanent updates. Indeed, I don't really remember this for the Leaps, they most of the time have several older versions still in.
Ah. Well perhaps you've never used VAX VMS ...
I did. But the multi-version-filesystem isn't something I'm really missing.
So sorry, its not just snapper that annoyed the hell out of me, even after much configuration, it was BtrFS as well.
Everyone is free to pick his favorites. I didn't want to convince you of anything, just wanted to point out that some issues can easily be solved..
Yes, getting rid of BtrFS wasn't easy since it was the RootFS, but it was do-able with just a little work since I was using LVM. -- A: Yes. > Q: Are you sure? >> A: Because it reverses the logical flow of conversation. >>> Q: Why is top posting frowned upon? -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
Wolfgang Mueller wrote:
I am going to install 42.3 on a computer whose hard disk I want to erase completely to change partitioning. This gives me the chance to choose the file system.
So far, I used ext4 only. But recently I read more about XFS and btrfs. But I know too little about them. Are they better than ext4? Or worse? Can you tell me something about their advantages and possible defects? Are there bugs?
XFS is reportedly well or better suited as a file system for many small objects. Unless you have specific requirements, ext4 is probably a safe choice. -- Per Jessen, Zürich (24.1°C) http://www.hostsuisse.com/ - dedicated server rental in Switzerland. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 26/08/2019 18.09, Wolfgang Mueller wrote:
I am going to install 42.3 on a computer whose hard disk I want to erase completely to change partitioning. This gives me the chance to choose the file system.
Why not 15.1? 42.3 is out of maintenance, you will have problems finding packages.
So far, I used ext4 only. But recently I read more about XFS and btrfs. But I know too little about them. Are they better than ext4? Or worse? Can you tell me something about their advantages and possible defects? Are there bugs?
XFS is certainly a very good filesystem for use on the /home or data partitions. For the root system, ext4 is probably better. Both are mature and stable. Btrfs is a very good and advanced filesystem, feature rich. I don't like it much, though. -- Cheers / Saludos, Carlos E. R. (from 15.0 x86_64 at Telcontar)
On 8/26/19 6:09 PM, Wolfgang Mueller wrote:
I am going to install 42.3 on a computer whose hard disk I want to
why that old version?
erase completely to change partitioning. This gives me the chance to choose the file system.
So far, I used ext4 only. But recently I read more about XFS and btrfs. But I know too little about them. Are they better than ext4? Or worse? Can you tell me something about their advantages and possible defects? Are there bugs?
I'm using ext4 everywhere, no problems. For btrfs, you need to reserve much more space for the partitions because of the subvolumes and snapshots (see the many threads on this mailing list already), and btrfs does not expose the real usable space to standard tools like 'df', so without using the btrfs-specific tools (and learning them), one might easily fall into the ENOSPACE trap. I can't tell much about bugs because I didn't follow closely, but btrfs showed up in the changes of the kernel pretty often in the early days. So at least if you really want to go with old 42.3, then this might be a problem. In newer openSUSE, I usually uninstall (and lock) the btrfs and snapper related packages, so they don't go into my way during updates: $ zypper locks | grep -E 'btrfs|snapper' 1 | btrfsmaintenance | package | (any) 2 | grub2-snapper-plugin | package | (any) 4 | snapper | package | (any) 5 | snapper-zypp-plugin | package | (any) 7 | yast2-snapper | package | (any) I have used XFS quite seldom, although it's known to be very stable, but why mess with ext3/4 which simply works? Well, in the end it's good that there are many file system types out there in the Linux world, and you can pick your choice to serve your personal needs. Have a nice day, Berny -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 08/26/2019 11:09 AM, Wolfgang Mueller wrote:
I am going to install 42.3 on a computer whose hard disk I want to erase completely to change partitioning. This gives me the chance to choose the file system.
So far, I used ext4 only. But recently I read more about XFS and btrfs. But I know too little about them. Are they better than ext4? Or worse? Can you tell me something about their advantages and possible defects? Are there bugs?
Best regards, Wolfgang
If it ain't broke, don't fix it! I haven't suffered a single filesystem corruption or file loss since moving to ext2 from reiserfs sometime around SuSE 10.0. Stay with ext4. -- David C. Rankin, J.D.,P.E. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On Mon, 26 Aug 2019 16:22:12 -0500 "David C. Rankin" <drankinatty@suddenlinkmail.com> wrote:
On 08/26/2019 11:09 AM, Wolfgang Mueller wrote:
I am going to install 42.3 on a computer whose hard disk I want to erase completely to change partitioning. This gives me the chance to choose the file system.
So far, I used ext4 only. But recently I read more about XFS and btrfs. But I know too little about them. Are they better than ext4? Or worse? Can you tell me something about their advantages and possible defects? Are there bugs?
Best regards, Wolfgang
If it ain't broke, don't fix it! I haven't suffered a single filesystem corruption or file loss since moving to ext2 from reiserfs sometime around SuSE 10.0. Stay with ext4.
I haven't moved from reiserfs and I haven't suffered a problem either, although I do use XFS as well, and my root is now btrfs. The one that gave me grief in the day was ZFS. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 08/26/2019 04:30 PM, Dave Howorth wrote:
On Mon, 26 Aug 2019 16:22:12 -0500 "David C. Rankin" <drankinatty@suddenlinkmail.com> wrote:
On 08/26/2019 11:09 AM, Wolfgang Mueller wrote:
I am going to install 42.3 on a computer whose hard disk I want to erase completely to change partitioning. This gives me the chance to choose the file system.
So far, I used ext4 only. But recently I read more about XFS and btrfs. But I know too little about them. Are they better than ext4? Or worse? Can you tell me something about their advantages and possible defects? Are there bugs?
Best regards, Wolfgang
If it ain't broke, don't fix it! I haven't suffered a single filesystem corruption or file loss since moving to ext2 from reiserfs sometime around SuSE 10.0. Stay with ext4.
I haven't moved from reiserfs and I haven't suffered a problem either, although I do use XFS as well, and my root is now btrfs. The one that gave me grief in the day was ZFS.
Chuckling... Yep, I still have reiserfs running along with ext3 and ext4 (on a 10 year old drive hanging off one box): mount /dev/sdc5 on / type ext3 (rw,relatime,data=ordered) /dev/sdc7 on /home type ext4 (rw,relatime,data=ordered) /dev/sda2 on /mnt/pv type reiserfs (rw,relatime) cat /mnt/pv/etc/SuSE-release SUSE LINUX 10.0 (i586) OSS VERSION = 10.0 (I bet it would still boot too :) -- David C. Rankin, J.D.,P.E. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 26/08/2019 17:42, David C. Rankin wrote:
On 08/26/2019 04:30 PM, Dave Howorth wrote:
On Mon, 26 Aug 2019 16:22:12 -0500 "David C. Rankin" <drankinatty@suddenlinkmail.com> wrote:
On 08/26/2019 11:09 AM, Wolfgang Mueller wrote:
I am going to install 42.3 on a computer whose hard disk I want to erase completely to change partitioning. This gives me the chance to choose the file system.
So far, I used ext4 only. But recently I read more about XFS and btrfs. But I know too little about them. Are they better than ext4? Or worse? Can you tell me something about their advantages and possible defects? Are there bugs?
Best regards, Wolfgang
If it ain't broke, don't fix it! I haven't suffered a single filesystem corruption or file loss since moving to ext2 from reiserfs sometime around SuSE 10.0. Stay with ext4.
I haven't moved from reiserfs and I haven't suffered a problem either, although I do use XFS as well, and my root is now btrfs. The one that gave me grief in the day was ZFS.
Chuckling...
Yep, I still have reiserfs running along with ext3 and ext4 (on a 10 year old drive hanging off one box):
mount /dev/sdc5 on / type ext3 (rw,relatime,data=ordered) /dev/sdc7 on /home type ext4 (rw,relatime,data=ordered) /dev/sda2 on /mnt/pv type reiserfs (rw,relatime)
cat /mnt/pv/etc/SuSE-release SUSE LINUX 10.0 (i586) OSS VERSION = 10.0
There's only one things wrong with the ReiserFS as distributed: it is single threaded. In the above config that's not a problem, but I had about 10 file systems running ReiserFS and the performance was awful since it was single threaded: no simultaneity, too much 'locking'. of course problems like that are fixed in Reiser4FS, but for a variety of reasons that isn't being released. The downside of ext4 is that it has to maintain backwards compatibility :-( It tries to be a B-tree file systems but isn't really, and it has so many options! It's a prime example of what I've come to terms 'coder masturbation'. The whole point of ResiserFS was its simplicity. It is a PURE B-tree file system. OK, so it gets to be a bit to simple and ends up single threaded :-( https://en.wikipedia.org/wiki/Reiser4 Reiser4 is a computer file system, successor to the ReiserFS file system, developed from scratch by Namesys and sponsored by DARPA as well as Linspire. Reiser4 was named after its former lead developer Hans Reiser. As of 2019, the Reiser4 patch set is still being maintained, but according to Phoronix, it is unlikely to be merged into mainline Linux without corporate backing. -- A: Yes. > Q: Are you sure? >> A: Because it reverses the logical flow of conversation. >>> Q: Why is top posting frowned upon? -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 08/29/2019 09:23 AM, Anton Aylward wrote:
https://en.wikipedia.org/wiki/Reiser4 Reiser4 is a computer file system, successor to the ReiserFS file system, developed from scratch by Namesys and sponsored by DARPA as well as Linspire. Reiser4 was named after its former lead developer Hans Reiser. As of 2019, the Reiser4 patch set is still being maintained, but according to Phoronix, it is unlikely to be merged into mainline Linux without corporate backing.
Bummer... I'm not sure I understand:
but according to Phoronix, it is unlikely to be merged into mainline Linux without corporate backing.
"corporate backing?" for kernel inclusion?? I guess they mean somebody to sign up to maintain the Reiser4 kernel code as changes to the kernel impact that part?? Let me know if you know what that means. I'm sure there is a reason that makes sense, but whatever it is, it's not something intuitive to me. -- David C. Rankin, J.D.,P.E. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 30/08/2019 07.43, David C. Rankin wrote:
On 08/29/2019 09:23 AM, Anton Aylward wrote:
https://en.wikipedia.org/wiki/Reiser4 Reiser4 is a computer file system, successor to the ReiserFS file system, developed from scratch by Namesys and sponsored by DARPA as well as Linspire. Reiser4 was named after its former lead developer Hans Reiser. As of 2019, the Reiser4 patch set is still being maintained, but according to Phoronix, it is unlikely to be merged into mainline Linux without corporate backing.
Bummer...
I'm not sure I understand:
but according to Phoronix, it is unlikely to be merged into mainline Linux without corporate backing.
"corporate backing?" for kernel inclusion??
Well, reiserfs 2 was backed by SUSE first, before the kernel guys accepted it.
I guess they mean somebody to sign up to maintain the Reiser4 kernel code as changes to the kernel impact that part?? Let me know if you know what that means. I'm sure there is a reason that makes sense, but whatever it is, it's not something intuitive to me.
-- Cheers / Saludos, Carlos E. R. (from 15.0 x86_64 at Telcontar)
David C. Rankin wrote:
Bummer...
I'm not sure I understand:
but according to Phoronix, it is unlikely to be merged into mainline Linux without corporate backing.
"corporate backing?" for kernel inclusion??
I'd understand that as 'we want a full-time payed developer responsible for such a module' I'd love to see it. We used it 10 years ago on our data acquisition system, and it was running circles around any other filesystem available (for Linux) then.... -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On Mon, Aug 26, 2019 at 06:09:51PM +0200, Wolfgang Mueller wrote:
So far, I used ext4 only. But recently I read more about XFS and btrfs. But I know too little about them. Are they better than ext4? Or worse? Can you tell me something about their advantages and possible defects? Are there bugs?
File system preferences depend a lot on use cases. We're using XFS for /home on 50 machines or so because of some testing I did a while ago (cough 12 years cough) where it turned out its performance was clearly superior. The scenario I've tested was multiple large files (hundreds of GB) being appended to from several NFS clients. This result isn't very surprising given that XFS was designed with video processing in mind. A. -- Ansgar Esztermann Sysadmin Dep. Theoretical and Computational Biophysics http://www.mpibpc.mpg.de/grubmueller/esztermann
Le 27/08/2019 à 10:48, Ansgar Esztermann-Kirchner a écrit :
(cough 12 years cough) where it turned out its performance was clearly
well, all of this is visibly obsolete, all may have changed since (on either direction :-() jdd -- http://dodin.org -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 27/08/2019 11.16, jdd@dodin.org wrote:
Le 27/08/2019 à 10:48, Ansgar Esztermann-Kirchner a écrit :
(cough 12 years cough) where it turned out its performance was clearly
well, all of this is visibly obsolete, all may have changed since (on either direction :-()
Tests over the years say that XFS excels on large file performance. -- Cheers / Saludos, Carlos E. R. (from 15.0 x86_64 at Telcontar)
Am 27.08.19 um 12:43 schrieb Carlos E. R.:
On 27/08/2019 11.16, jdd@dodin.org wrote:
Le 27/08/2019 à 10:48, Ansgar Esztermann-Kirchner a écrit :
(cough 12 years cough) where it turned out its performance was clearly
well, all of this is visibly obsolete, all may have changed since (on either direction :-()
Tests over the years say that XFS excels on large file performance.
What is a large file? Approx half of my files are around 10MB, the other half around 30 to 40 MB. Those are image files and I think they are quite large. Then I have countless really small files (some KB), and some hundreds of videos at approx. 1 GB each. Now, do I need a system for small files or for large files? Currently I use ext4, except on external disks where I have reiserfs Oh: and all disk are LUKS encrypted. Does the file system have an influence on that? -- Daniel Bauer photographer Basel Málaga https://www.patreon.com/danielbauer https://www.daniel-bauer.com -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 27/08/2019 13.00, Daniel Bauer wrote:
Am 27.08.19 um 12:43 schrieb Carlos E. R.:
On 27/08/2019 11.16, jdd@dodin.org wrote:
Le 27/08/2019 à 10:48, Ansgar Esztermann-Kirchner a écrit :
(cough 12 years cough) where it turned out its performance was clearly
well, all of this is visibly obsolete, all may have changed since (on either direction :-()
Tests over the years say that XFS excels on large file performance.
What is a large file?
Video :-) Rather very large, but not huge. For example, delete a video file in XFS is instantaneous. XFS is excellent on video streaming, that is continuosly creating and deleting large files. You can think large in the context of hundreds or thousands of file blocks that have to be indexed. Large inodes or equivalent.
Approx half of my files are around 10MB, the other half around 30 to 40 MB. Those are image files and I think they are quite large.
They are.
Then I have countless really small files (some KB), and some hundreds of videos at approx. 1 GB each.
Now, do I need a system for small files or for large files?
XFS is fine for that, although it is not as good as reiserfs for small files.
Currently I use ext4, except on external disks where I have reiserfs
I'm phasing out my reiserfs partitions. :-( Now I only have one, but it does several services.
Oh: and all disk are LUKS encrypted. Does the file system have an influence on that?
I don't think so. Encryption does not change file block size, I believe. -- Cheers / Saludos, Carlos E. R. (from 15.0 x86_64 at Telcontar)
Le 27/08/2019 à 12:43, Carlos E. R. a écrit :
On 27/08/2019 11.16, jdd@dodin.org wrote:
Le 27/08/2019 à 10:48, Ansgar Esztermann-Kirchner a écrit :
(cough 12 years cough) where it turned out its performance was clearly
well, all of this is visibly obsolete, all may have changed since (on either direction :-()
Tests over the years say that XFS excels on large file performance.
and also on many small files :-) only drawback I know of is that one can't reduce the file system size (in case /home is too large versus /, for example) jdd -- http://dodin.org -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On Tue, 27 Aug 2019 13:02:57 +0200 "jdd@dodin.org" <jdd@dodin.org> wrote:
Le 27/08/2019 à 12:43, Carlos E. R. a écrit :
On 27/08/2019 11.16, jdd@dodin.org wrote:
Le 27/08/2019 à 10:48, Ansgar Esztermann-Kirchner a écrit :
(cough 12 years cough) where it turned out its performance was clearly
well, all of this is visibly obsolete, all may have changed since (on either direction :-()
Tests over the years say that XFS excels on large file performance.
and also on many small files :-)
It's quite slow on metadata operations on small files (well, large directories actually). Doing an ls or an rm on a directory of a million files can take quite a while.
only drawback I know of is that one can't reduce the file system size (in case /home is too large versus /, for example)
jdd
-- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
Ansgar Esztermann-Kirchner wrote:
File system preferences depend a lot on use cases. We're using XFS for /home on 50 machines or so because of some testing I did a while ago (cough 12 years cough) where it turned out its performance was clearly superior. The scenario I've tested was multiple large files (hundreds of GB) being appended to from several NFS clients.
We did a similar choice (even more than 12y ago). And while I still like (and use) XFS for data partitions with large files I have to admit that this (XFS via NFS mount) was the only one to give me a total and unrecoverable data loss... -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 30/08/2019 11.18, Peter Suetterlin wrote:
Ansgar Esztermann-Kirchner wrote:
File system preferences depend a lot on use cases. We're using XFS for /home on 50 machines or so because of some testing I did a while ago (cough 12 years cough) where it turned out its performance was clearly superior. The scenario I've tested was multiple large files (hundreds of GB) being appended to from several NFS clients.
We did a similar choice (even more than 12y ago). And while I still like (and use) XFS for data partitions with large files I have to admit that this (XFS via NFS mount) was the only one to give me a total and unrecoverable data loss...
I had total losses with reiserfs, xfs, and ext3. And in the distant past, fat. I also experienced a reproducible total loss event with btrfs. With ntfs... nearly so, but a proprietary (paid) tool managed to recover all the important files. -- Cheers / Saludos, Carlos E. R. (from 15.0 x86_64 at Telcontar)
On Friday, 30 August 2019 19:29:35 ACST Carlos E. R. wrote:
On 30/08/2019 11.18, Peter Suetterlin wrote:
Ansgar Esztermann-Kirchner wrote:
File system preferences depend a lot on use cases. We're using XFS for /home on 50 machines or so because of some testing I did a while ago (cough 12 years cough) where it turned out its performance was clearly superior. The scenario I've tested was multiple large files (hundreds of GB) being appended to from several NFS clients.
We did a similar choice (even more than 12y ago). And while I still like (and use) XFS for data partitions with large files I have to admit that this (XFS via NFS mount) was the only one to give me a total and unrecoverable data loss...
I had total losses with reiserfs, xfs, and ext3. And in the distant past, fat. I also experienced a reproducible total loss event with btrfs. With ntfs... nearly so, but a proprietary (paid) tool managed to recover all the important files.
The only total losses I've had have been with LVM (many years ago, back in the days of Fedora Core 4) and btrfs. For the last 7 years or so all my critical data storage has been on RAID10/ext4 (with nightly backups to an external drive) using ext4. -- ============================================================== Rodney Baker VK5ZTV rodney.baker@iinet.net.au CCNA #CSCO12880208 ============================================================== -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
participants (14)
-
Andrei Borzenkov -
Ansgar Esztermann-Kirchner -
Anton Aylward -
Bernhard Voelker -
Carlos E. R. -
Daniel Bauer -
Dave Howorth -
David C. Rankin -
jdd＠dodin.org -
Knurpht-openSUSE -
Per Jessen -
Peter Suetterlin -
Rodney Baker -
Wolfgang Mueller