vinícius mota wrote:
Briefly, the problem is that the copy speed between the partition in
which I am running my Fedora and another partition where I store backup
data is now 50K/s, which is extremely slow by any standart (it has been
usually around 10M/s, almost 200 times faster).
Wow, that's slow between partitions.
One of the disks is partitioned between two operational
systems, Fedora 14 64 and Windows 7 64, and the other disk contains the
above referred partition where I store my backups (I call it backup
disk).
I login into the windows and
made a copy of some AVI files (3G approx) to the backup disk, which went
normally
So hardware is ok and, also, Windows' disk handlers are probably ok.
I move data within the backup iself disk using windows also without
any
problem.
I don't think this would move the files, would it? Wouldn't the
directory entries, only, change? Could you try a 'copy' within the
backup (under windows), just to be sure. Preferably something quite
large.
I try therefore to make the inverse process, a copy from the
backup disk into my home folder in Fedora. It went normally, copying 4 G
also within few seconds, actually with an excellent speed of 50M/s.
Good
[Google might show] one cause could be that the source file system
(in my
case the Fedora home folder) would be near its full capacity.
I wouldn't have thought so; to me it looks much more like a data
problem on the destination, or a problem in the F14 disk handlers (but
would have been quite widely reported, by now?).
It is clear that you have been using Fedora for some time so I expect
you have already done what I will ask, but you did not say. My first
thought on reading the symptoms was that, possibly, the backup
partition is getting full with 'wastebasket/recyclebin' files, which
have to be 'deleted for real' before the 'new files' that you are
copying can be copied. And, maybe, Windows is a little bit faster at
that than F14 NTFS handlers (maybe). Maybe F14 is 'deleting for real'
at a single sector at a time, or something?
I wonder, also, if the number of IO reads and writes on the backup
disk could be measured and reported; that might provide a clue as to
whether there is excessive processing, or whether there is otherwise
some software delay. But I do not know how to measure IO transactions.
He is the line of my /etc/fstab file which defines my backup disk, with
the label common
LABEL=common /home/vinicius/Desktop/common ntfs
auto,user,sync 0 0
Am I correct in thinking that 'sync' disables write-caching? It might
be interesting to change that, and see if performance is still just as
bad.
Best of luck, Ron