Slashdot Mirror
Best Shrinkable ReiserFS Replacement?
paulkoan writes "I have been using ReiserFS for my file system across a few servers for some time now (follow the link below for details of my experience). I can't foresee the future of ReiserFS, but if I'm going to have to migrate as support diminishes, I'd like to begin that process now. My criteria are: in-kernel support, shrinkable, and has good recovery when the file system is not closed properly. That shrinkable requirement precludes a lot of options. What's a good replacement for ReiserFS?"
I initially chose ReiserFS because I was building a MythTV system and it was the recommended FS across the board, from small to large files. I've had good experiences with ReiserFS and it has had a pummeling. That MythTV box for example has a very volatile environment and loses power on a regular basis. I haven't lost any data through any of these outages.
Compare this to my brief foray into XFS on the same box, where 25% of the filesystem ended up in lost+found with numbers for filenames. When this happened a second time on a different system I decided XFS wasn't for me — and I really don't get the point of a journalled filesystem that will keep data relatively safe, but then remove any means to identify it when things go wrong.
But everyone has good and bad experiences with filesystems, ReiserFS included. XFS has a good rep, my experience aside.
I initially chose ReiserFS because I was building a MythTV system and it was the recommended FS across the board, from small to large files. I've had good experiences with ReiserFS and it has had a pummeling. That MythTV box for example has a very volatile environment and loses power on a regular basis. I haven't lost any data through any of these outages.
Compare this to my brief foray into XFS on the same box, where 25% of the filesystem ended up in lost+found with numbers for filenames. When this happened a second time on a different system I decided XFS wasn't for me — and I really don't get the point of a journalled filesystem that will keep data relatively safe, but then remove any means to identify it when things go wrong.
But everyone has good and bad experiences with filesystems, ReiserFS included. XFS has a good rep, my experience aside.
ReiserFS works. It is merged with the mainline kernel trunk so it will be able to secure enough man power to at least avoid bit-rot and incompatibility to future kernel versions. You don't have to worry about suddenly losing your files now Hans isn't involved in the project, some kernel modules have gone for years without an update and still work. I doubt that this will even become one of them since so many people are using this file system and lets face it, it is a good file system nomatter who wrote it (lets not forget he was a known arsehole before he killed his wife and it didn't matter then).
The worst thing that could happen is ReiserFS slowly falling into disuse and becoming deprecated in three or four years, you will have plenty of time to worry about this later, just take a deep breath and put down your file system tools, this will all be OK.
When Argumentum ad Hominem falls short, try Argumentum ad Matrem
Don't fix it. Reiser3 is in the mainline kernel. Why bother messing with your working (and apparently robust) system?
General Relativity: Space-time tells matter where to go; Matter tells space-time what shape to be.
ZFS isn't available on Linux. It is a great enterprise class file system, but for the type of application discussed here it is probably overkill in terms of management complexity, etc. Not that it would be bad or wrong to use, and you could benefit from some of the features of course, but it really shines in demanding environments. In any case unless you're running openSolaris it isn't an option.
Ext3 in my experience is just plain inferior to ReiserFS. Recovery and formatting are both slow as death. Like the OP I have yet to suffer any data loss on a ReiserFS since way back in the early days when it first came out. Ext3 seems pretty reliable as well, but the slow recovery times are annoying and once in a while it seems like a whole filesystem just plain becomes irretrievably corrupted. OTOH it does demand less CPU overhead. Rarely a BIG issue, but can be with HTPC type systems.
Overall though I don't think you have a lot of choice. XFS or JFS might be perfectly good solutions, not really had a need to mess with either of them myself so I can't comment. Obviously ReiserFS looks like it has about reached the end and that pretty much leaves Ext3 as the only man left standing in the ring at this point. Cheer up, it works well enough, you'll just have to live without the shrink functionality... ;).
"Malo periculosam, libertatem quam quietam servitutem." -- Jefferson
ReiserFS is OSS, if the project leaders are taking it in a direction people don't like, then fork it.
I would stick with ext3 - it is really the only option that meets your needs (which is why I'm using it as well). Note that I'd avoid using LVM - there is some kind of bug in some versions of LVM that causes massive data loss in some very rare circumstances. I recently lost a few hundred GB of data on a RAID due to this issue. (Google for "access beyond end of device lvm".) Ran fsck to clean up some errors after a crash while in RAID recovery mode and suddenly I had massive data loss on an entirely different lvm logical volume - it was obvious that the fsck somehow crossed the logical volume boundaries which should not be possible.
In the end I ended up restoring critical data from backups (which did not include mythtv recordings), and watched what remained of my recordings (complete with 10 second patches of video jumping between shows). I had to completely wipe out everything on the raid and start over. I no longer run lvm - I used to swear by it but it will be a while before I go back to it. My few non-lvm partitions (root, boot) had no issues at all even though they were subject to the same treatment.
How is the parent offtopic? Funny perhaps (at least I thought it was funny) but definitely ON topic of shrinkable filesystems.
"The agriculture ministry is not in charge of Gundam" - Japanese ministry official.
Originally, XFS lacked support for the equivalent to write:ordered in ext3. In practice that meant that while the metadata would be intact, your files (and thereby directories) might be blobs of zeros/gone. That is probably what happened to him, and it is what happened to me when I used it long ago. I think I have read that XFS has gotten this capability later. For the topic, I'd recommend ext3 too. It's proven, can do the online grow/offline shrink that is so handy, has 3 different settings for speed/reliance, and is well supported. But I am not a sysadmin.
Religion is regarded by the common people as true, by the wise as false, and by rulers as useful.
ReiserFS is still maintained quite well, and Reiser4 is coming along well. I personally use ReiserFS exclusively, and recommend sticking with it, or possibly trying out Reiser4.
Is it broke for you? If not, don't fix it.
Need an automatic screenshot taker? Try here.
Am I missing something... Isn't one of the advantages of Open Source is the ability to maintain and improve a product no matter what happens to any particular person or company?
Perhaps the Open Source Community should have project safety rating for an open source project. Like a risk level. So say without the person the project is doomed to die. To fully supported so if any one comes or go the product will continue.
If something is so important that you feel the need to post it on the internet... It probably isn't that important.
Most people are using Linux for HTPC applications, BSD is a whole other kettle of fish...
I'm not saying ZFS is HARD to admin, just that a basic ext3 fs is nothing at all to admin. If you know ZFS, it is no big deal. For your average garden variety user they will never take advantage of the ZFS features anyway, so they'd be better of just going with ext3. There is a lot better chance their recovery tools support it, etc.
So, I'll agree with you, if the user happens to be sophisticated and using BSD, then go for ZFS, why not?
"Malo periculosam, libertatem quam quietam servitutem." -- Jefferson
> Reiser has a codebase of an insane
wc -l says otherwise (2.6.25.9):
"Between strong and weak, between rich and poor [...], it is freedom which oppresses and the law which sets free"
ReiserFS was a great filesystem. Certainly isn't a WORSE one now than it was before in and of itself. It is just a matter of ongoing support. Whatever the reason people are dropping support and inevitably an unmaintained file system is going to become at best a marginal legacy tool. Given that it isn't even a default supported Linux fs chances are it will be broken as of kernel X, and then you will be pretty much forced to migrate, so why bother with it?
As to WHY, that I have to assume is political, there was never anything technically wrong with it. In fact there was pretty much everything technically RIGHT about it... Reading between the lines my hunch would be
"Malo periculosam, libertatem quam quietam servitutem." -- Jefferson
I assume the "shrink" requirement is to preclude discussion of the most viable alternatives.
Wouldn't leaving ReiserFS because it will supposedly become unsupported at some point in the future become self-fulfilling?
After all, who is going to fix it if nobody uses it? And like everything else, it will need to be fixed at some point.
If you find it to be a good filesystem, I say keep using it. If it is a good filesystem, someone will maintain it. If not, then it dies because it deserves to and not because of FUD.
blah blah blah
...in many production environments except for emergencies. There's no way I can take a 3TB production filesystem offline to do a dump/restore (or even a restore) just to shrink a filesystem.
There are times we have to shrink one filesystem (volume, whatever) so we can grow another with the available space, without halting dozens of engineers for hours or a day. Lots of people have this same problem.
With something like a NetApp, I can grow and shrink a filesystem at will. Quick, simple, painless. I realize we are talking free software and commodoty or similar hardware vs several hundred thousand dollars worth of proprietary HW & SW, but the point is that it's doable, and there are reasons to do it.
We have NetApps for or tier 1 space, and they work great. We use Linux on Supermicros and Dells for tier 2, and desperately need easier disk management there.
No, it was awful in production use. Any disk failure on a RAID set, as can be expected to occur in any large environment, would lead to a confused ReiserFS whose own recovery tools would destroy it and which could no longer be backed up reliably. This has happened with no other Linux compatible filesyste I've ever seen: the others simply report a failed access, they do not lock when trying to read the ruined files.
ReiserFS was only suitable for high-speed, random access, restorable via other means filesystems such as NNTP servers and web proxies. For root partitions, boot partitions, and home directories, it was unstable and dangerous to use.
If you find it to be a good filesystem, I say keep using it. If it is a good filesystem, someone will maintain it. If not, then it dies because it deserves to and not because of FUD.
.
Hope is not a plan.
It is reasonable to wary of any project so uniquely identified with a single individual.
But a filing system? How many developers have the freedom and the confidence to take on something so elemental?
Prison isn't about getting to do the things you love with free room and board.
Nor is it about harming society by preventing criminals from making meaningful contributions, just to spite them.
Visual IRC: Fast. Powerful. Free.