Request for comments re filesystem types ext4, XFS and reiser
in [Suse]
|
Prev: 2Wire help needed.
Next: another problem with suse 11.2
From: Roger Pryor on 13 Dec 2009 14:27 Hello; I have been using the Reiser filesystem for several years now, since I had data loss problems with power loss crashes under Ext2 and early versions of Ext3. How are people finding Ext4 for data loss when their machine loses power? I'm planning a shift to OpenSUSE 11.2 early next year, and I understand that Ext4 is the default and and am thinking of shifting. I have used XFS in the past for very large files (4 GB), but encountered a different problem in that re-use of a filespace seemed to invoke multiple track 0 seeks, with the resulting extreme loss of performance. This could be a hardware problem, but I'm not sure. In case it matters, I have a mixture of very small and very large (4 GB) files. TIA, and Seasons Greetings to all. -------------------------------------------------------------------------- Roger Pryor Email: rjpryor(a)infoserve.net Sunset Crest, Barbados, W.I.
From: Kevin Nathan on 15 Dec 2009 18:52 On Sun, 13 Dec 2009 15:27:15 -0400 Roger Pryor <rpryor(a)infoserve.net> wrote: >Hello; > >I have been using the Reiser filesystem for several years now, since I >had data loss problems with power loss crashes under Ext2 and early >versions of Ext3. I abandoned ReiserFS a few years ago when it refused to mount a 20GB partition with two bad sectors at the end of the partition. There was only 6 or 8GB of data at the beginning of the partition, but I could not recover *any* of it. POS filesystem as far as I am concerned! :-) >How are people finding Ext4 for data loss when their >machine loses power? No power loss, yet, so can't say for sure. >I'm planning a shift to OpenSUSE 11.2 early next >year, and I understand that Ext4 is the default and and am thinking of >shifting. Yes, it's the default but you can use whatever you prefer. I went with ext2 for the /boot partition and ext4 for / (20GB), /home (80GB), /var (30GB), /tmp (30GB) and XFS for /local (200GB). In a few weeks, /home will get its own 400 GB drive and I will probably go to XFS for it then. >I have used XFS in the past for very large files (4 GB), but >encountered a different problem in that re-use of a filespace seemed to >invoke multiple track 0 seeks, with the resulting extreme loss of >performance. This could be a hardware problem, but I'm not sure. In >case it matters, I have a mixture of very small and very large (4 GB) >files. > We use XFS on all our servers and the response time really improved when we switched. It's mostly small to medium files with just a few very large files. There have been a few power outages at the co-location over the last few years and we haven't had any data loss with XFS... :-) -- Kevin Nathan (Arizona, USA) Linux Potpourri and a.o.l.s. FAQ -- (temporarily offline) Open standards. Open source. Open minds. The command line is the front line. Linux 2.6.31.5-0.1-default 15:45pm up 1 day 1:04, 13 users, load average: 0.00, 0.03, 0.00
From: Ulick Magee on 15 Dec 2009 19:02 Roger Pryor wrote: > > I have been using the Reiser filesystem for several years now, since I > had data loss problems with power loss crashes under Ext2 and early > versions of Ext3. I have ext3 on my 11.1 laptop; sometimes it locks solid some time after resuming from suspend and the only way out is to power off. fsck is then forced on restart, but I've never had any problems, and it usually only delays the boot process by a few seconds. Incidentally, since going from 11.0 to 11.1 the periodic and lengthy force-fscking of ext3 partitions seems to have gone away... -- Ulick Magee Free software and free formats for free information for free people. Open Office for Windows/OSX/Linux: http://www.openoffice.org openSUSE Linux: http://en.opensuse.org
From: Chris Cox on 15 Dec 2009 19:47 On Sun, 2009-12-13 at 15:27 -0400, Roger Pryor wrote: > Hello; > > I have been using the Reiser filesystem for several years now, since I > had data loss problems with power loss crashes under Ext2 and early > versions of Ext3. How are people finding Ext4 for data loss when their > machine loses power? I'm planning a shift to OpenSUSE 11.2 early next > year, and I understand that Ext4 is the default and and am thinking of > shifting. I have used XFS in the past for very large files (4 GB), but > encountered a different problem in that re-use of a filespace seemed to > invoke multiple track 0 seeks, with the resulting extreme loss of > performance. This could be a hardware problem, but I'm not sure. In case > it matters, I have a mixture of very small and very large (4 GB) files. > Right now, I don't see a real advantage to ext4. It's basically an "almost reiserfs". Certainly a step up from ext3 with regards to enterprise features, but slower in many benchmarks. Ultimately, I'm banking on nilfs2 or brtfs... but that will be a year or more away still. There might be some others that are up and coming as well. Ext4 set its goals very low and achieved them (well sort of). The major problem with reiserfs is simply with regards to maintainers for it. Apart from that, it's still the only good enterprise level fs choice. It's sort of sad (with regards to reiserfs). Similar to the demise of KDE3. If somebody forced me to move away from reiserfs, I'd choose XFS today. But I really don't even want to consider that... I'd rather wait for one of the future filesystems to get deployed. With that said, I did let openSUSE 11.2 use ext4... that's what it wanted to use... so I'll let it (and I may live to regret that).
From: David Bolt on 16 Dec 2009 02:08
On Tuesday 15 Dec 2009 23:52, while playing with a tin of spray paint, Kevin Nathan painted this mural: > On Sun, 13 Dec 2009 15:27:15 -0400 > Roger Pryor <rpryor(a)infoserve.net> wrote: > >>Hello; >> >>I have been using the Reiser filesystem for several years now, since I >>had data loss problems with power loss crashes under Ext2 and early >>versions of Ext3. > > I abandoned ReiserFS a few years ago when it refused to mount a 20GB > partition with two bad sectors at the end of the partition. There was > only 6 or 8GB of data at the beginning of the partition, but I could not > recover *any* of it. POS filesystem as far as I am concerned! :-) I haven't had that issue, but I have had one issue with a reiserfs file system being unrecoverable. That was quite a long time ago, one of the 8.x releases IIRC, but nothing since. I still use it for file systems where I know I'm going to have large quantities of small files. >>How are people finding Ext4 for data loss when their >>machine loses power? > > No power loss, yet, so can't say for sure. I'd expect it to be virtually as stable as ext3, if not more so, but I've only got it in use on a single system and that one hasn't had a lot of use lately so can't make any comments as yet. >>I'm planning a shift to OpenSUSE 11.2 early next >>year, and I understand that Ext4 is the default and and am thinking of >>shifting. > > Yes, it's the default but you can use whatever you prefer. I went with > ext2 for the /boot partition and ext4 for / (20GB), /home (80GB), /var > (30GB), /tmp (30GB) and XFS for /local (200GB). In a few weeks, /home > will get its own 400 GB drive and I will probably go to XFS for it > then. The one I'm using now uses a nice mix of ext2, ext3, reiserfs and xfs. As I've done for many years, I use ext2 for /boot. On this system I use ext3 for / , /tmp , /usr , /var/log and a couple of partitions mounted under /local . I use reiserfs for /srv and /srv/obs , which is a separate 256GB partition used for my own local build service. The xfs file systems are /home and a few more file systems also mounted under /local . >>I have used XFS in the past for very large files (4 GB), but >>encountered a different problem in that re-use of a filespace seemed to >>invoke multiple track 0 seeks, with the resulting extreme loss of >>performance. This could be a hardware problem, but I'm not sure. The smart tools are your friend. Set it off doing a long test and then see what the results are when it finishes. Expect it to take a long time to do so. When doing a test on my 500GB drives, it gives an estimate of 137 minutes to finish, and 255 minutes to finish testing my 1TB drive. This isn't really surprising since it does a full surface scan. >>In >>case it matters, I have a mixture of very small and very large (4 GB) >>files. >> > > We use XFS on all our servers and the response time really improved > when we switched. It's mostly small to medium files with just a few > very large files. There have been a few power outages at the co-location > over the last few years and we haven't had any data loss with XFS... :-) I have had exactly one data loss under xfs caused by a power loss, and it was within the last few months. Luckily, the loss was an ISO image that had just finished being created when the power failed and so I was able to recreate it with little issue. Still, it was a bit disconcerting to find a 4GB file listed as 0 bytes. Regards, David Bolt -- Team Acorn: www.distributed.net OGR-NG @ ~100Mnodes RC5-72 @ ~1Mkeys/s openSUSE 11.0 32b | | openSUSE 11.2 32b | openSUSE 11.0 64b | openSUSE 11.1 64b | openSUSE 11.2 64b | TOS 4.02 | openSUSE 11.1 PPC | RISC OS 4.02 | RISC OS 3.11 |