Rajko M. wrote:
On Friday 14 September 2007 00:12, Aaron Kulkis wrote:
Hans Witvliet wrote:
....
In my experience, reiserfs is not safe. It frequently fails
(corruption and/or file loss) in the event of improper shutdown
such as power failure or system crash.

Conversely, in Iraq, where I experienced frequent power-related
shutdowns, and a couple crashes (due to I think, running out of
swap space), I never lost a single file using ext3 and xfs.
I could be mistaken, but i thought the use of xfs inherrently required
the use of a no-break power supply, as it holds more data in mem,,,,,
Both are journalled filesystems, but there seems to be an
implementation error in Reiserfs 3.x, because I had several
fsck failures after power outages.

XFS's journalling mechanism seems (in my experience) to be
much closer to the ideal.  Other people have experienced
losses with XFS, but I'm beginning to believe that those
were hardware related (physical defects on the hard drive),
which currently, a problem to which currently all filesystems
are susceptable.

The link http://pages.cs.wisc.edu/~vijayan/vijayan-thesis.pdf is about serious attempt to evaluate files systems. It is long read, even if you read only conclusions, but it seems fair. You good experience with XFS is a matter of luck or type of work that you do. Having UPS is good for any computer if one wants to prevent data loss, but in case of XFS is even more important as it keeps big chunks of work in memory, and power loss can delete hours of work. File system will be fine, but you would have to rewrite all that wasn't saved and that can be a lot. You bad experience with reiserfs is probably partially due to fsck. That FS is not well suited to hold backup of another reiserfs, because if it has to do fsck, both main FS and backed up can be mixed up in one mess. I don't have handy article about this, but fsck.reiserfs is not what you want to do on regular basis. They tell that explicitely that is dangerous when you run fsck.
The only time I ever ran fsck.reiserfs was after crashes
(boot up would detect that the fs was "stale" and ran it
automatically...leaving unresolved errors, which would dump
the system into run-level S and force me to run fsck.reiserfs
again manually, until there were no more unresolved errors).

Perhaps the default behavior on Reiserfs was wrong?

Judging from what you wrote



--
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to