hi, just wanted to point-out that linux 2.4 also supports it out of the box...

This has never happened to me, thankfully, and I've been using ext3 for a couple of years now under different distros. When setting up a Slackware server a few weeks back I did get corruption under Reiser though, and had to reinstall. I decided to change to ext3. This also happened to a friend of mine on an external drive he uses to backup his data. It wasn't a Slackware issue as he doesn't use Slack, but he was pretty annoyed as there was a lot of useful data on there! I think I'm currently using Reiser under Ubuntu and Gentoo and it hasn't caused any problem there.

Some nice filesystem benchmarks (includes ext3 & reiserFS):

BENCHMARKING LINUX FILESYSTEMS

Decide for yourself.

Great link. Strange that Reiserv4 results are so low. And my favourite Ext2 has performed greatly.

Yes, there are a number of interesting results in that report...and surprises.

.

The most interesting for me were filesystem mount time and total free space diagrams.

Well, that certainly clears things up. JFS is the clear winner.

Is it? Some of the more important arguments are completely absent from this thread:

- Ext2/Ext3 provide good support for filesystem quota, reiserfs does not. Maybe not interesting for workstations, but it certainly is for servers. XFS does also provide good quota support.
- Ext2/Ext3 provide good support for extended attributes and access control lists (ACLs). If you wan't to go beyond 'ugo' file permissions (which can often be useful on servers), you'll definitely want ext2, ext3, or XFS.
- reiserfs has worse bad block handling (it can't do much with damaged metadata blocks).
- reiserfs can do ugly things when a full tree rebuild is done, and will try to find anything that looks like a part of a filesystem. You can guess what happens if you have a filesystem backup of a reiserfs filesystem on a reiserfs filesystem .

All and all it is not a bad filesystem, but it misses some serious features, especially for servers. XFS seems to be a more logical successor to ext3, providing the same nifty modern features like balanced trees, but providing quota, ACLs et al. at the same time.

This is not really the case. The installer just picks the last filesystem in the list as the default. This will make the filesystem of 'file system kernels' (e.g. xfs.s) the default. As far as I can see the bare.i filesystems are just ordered.

I don't understand why mounting a Reiser3 volume benchmarked so slow, I have an imap mail server with maildir structure on a PII 266, running Reiser3FS. When it is started, mounting the volume is very fast.

I don't think that the various filesystems have much real world performance difference unless the requirements are "unusual"

I'm not so sure. Per the official Slackbook

It is the default with bare.i, because it is at the bottom of the list. Besides that the book isn't written by Pat. Of course, this is just my guess after reading the setup script. But I think that there is little ground to say Pat recommends reiserfs because it is the default in the Slack setup with the bare.i kernel (and some other kernels).

Maybe the book isn't written by Pat, but Patrick links to Slackbook from the Slackware home page, so that's good enough for me

found this on debian-administration.org, seems interesting so i thought i'd post it:

Filesystems (ext3, reiser, xfs, jfs) comparison on Debian Etch

You can change the frequency of checks using tune2fs. You can even set it to never check at all, but that's pretty obviously playing with fire. I set mine to a higher number (you can specify a time interval or the number of mounts, or both) because I've been booting many times a day trying out different distros to try to help a friend with a badly corrupted Windows laptop find a Linux system he can use till he can afford an Apple machine.

It's very easy and only takes a few seconds, though the manual rightly cautions being careful when using tune2fs.