yoros@xxxxxxxxxx wrote:
Hi,
I have executed xfs_fsr and with a very good results (thank you).
My intention is to know why this situations may happend and why XFS is
slow deleting files and directories.
First, I must say that I have not found XFS (the current versions)
to be slow in delete at all. Try running FreeBSD once :-)
However, it is fundamentally true that a journalled file system
(such as XFS, EXT3, JFS, etc) will be slower in operations that
have large metadata impacts. That is, anything that affects the
structure on disk (where files are, where empty space is, etc)
will be written to the journal in case of a crash/failure before
all of the disk structures could be updated. This is how such
filesystems can be remounted in moments after a crash without an
extensive fsck pass to revalidate the file system structure.
EXT2 is *very* fast at deleting whole directory trees since it does
so all in its cache and then writes things out to disk as it gets
time. XFS is not that fast. But it is rather fast.
BTW - I have had slow open(O_CREAT|O_TRUNC) on EXT2 when the file
already existed and was large... XFS is very much faster at that
on my systems. (This was overwriting files in the 500+MB range)
--
Michael Sinz -- Director, Systems Engineering -- Worldgate Communications
A master's secrets are only as good as
the master's ability to explain them to others.
|