You may remember just recently I talked about XFS and its support in various Linux distributions (in the context of a multi-distro review).
Shortly thereafter, I switched entirely to Suse 10.1, and converted most of my hard drives (one for video, one for audio, one for documents) over to XFS. This involved swapping data out to free space on other drives and to my external hard drive. I kid you not when I say that transferring files back to the XFS-formatted drive took literally a day per drive. That’s right, transferring, say, 200GB of data took 24 hours.
I had chosen XFS because it was supposed to be a very fast filesystem, especially good for media collections. Well, it didn’t take me long to realize that something was terribly wrong, and to this day I’m not sure if it was the particular version of the XFS tools or what (I used both YaST’s partitioner and regular old
mkfs.xfs), but it had to go. I reformatted the drive again, this time as ext3 with default settings, and transferring that same approximately 200GB took about 90 minutes (and that only because 100 of those gigabytes had to go across a USB connection).
So, I’ve more or less given up on XFS, and it looks like I’m not the only one. Apparently, a bunch of Debian developers have decided they don’t like it, either.
I am through with XFS, once and for all. Well, at least for laptops. I still think it’s a good filesystem when you can ensure that the power never goes, and your hardware is reliable, but it’s just not adequate for laptops or even desktops.
I exactly had the same kind of problem on my workstation 2 years ago. A single hard reboot used to totally fuck up some random files (which were opened in read-only mode before the crash, like libraries) on my file system. Since that day I regularly spite on XFS, and I am quite amused that more than 2 years later, XFS is still totally crap!
Any damage? So far, only minor things – i.e. the Gaim buddy list… Nevertheless, it might be minor, but there is always _something_. And I don’t want to wait until massive fsckup happens. And, since I’m already using ext3 as it is more solid for my root partition, I think I will be switching back to it. Specially at my servers, where I still value more reliability over speed.
It certainly worked better than raiser. But as said, that old machine has been crashing, and when I copied over the data for my dad to his new machine, I noticed that his .gconf file was destroyed by XFS. All 0-bytes, typical for XFS. If the file is damaged, just fill it up with zeroes? I’ve seen 0-bytes a couple of times before in log files on XFS partitions that were open during a crash. It sucks that gconf didn’t keep a backup.
Of course, all these grievances are with XFS’s penchant for destroying data, but I think it’s safe to say that I won’t be going back to XFS as a filesystem unless something fundamentally changes—it’s just neither the speed demon it’s supposed to be or a viable choices in a world where power outages (for instance) occur more frequently than we’d like.
Goodbye, XFS! Hello, ext3 (and maybe ext4 someday)!