|
|
Jim Henderson wrote:
>>>> In fairness, when was the last time you tried copying a file that
>>>> somebody else was still using? Not so easy, is it?
>>> Trivial on Linux.
>> As long as you don't care about correctness. ;-)
>
> Depends a lot on a lot of different factors. Since you're backing up an
> inode, the inode is going to be generally valid (unless it's a special
> file of some sort), but whether it's the version that's being modified in
> memory on some computer somewhere over a network or not depends on when
> the user last saved the file.
No. I'm saying if the file is open, and you're trying to back it up while
it's changing, you're going to have problems with consistency. The file
system might be OK, but whatever you stat() from the inode might be wrong by
the time you're actually copying the data. Or if someone renames X to Y, you
might wind up backing up neither.
"tar: file changed as we read it"
And backing up a SQL database file while SQL is working with it? Not a good
idea. Such a bad idea that SQL servers tend to have alternate mechanisms to
make it work. :-)
Of course, backing up *any* open file is going to be problematic, unless
everyone uses transactions in their file system (and I don't mean just
journaling, but real SQL-like transactions with multiple changes and commits
and rollbacks and such).
--
Darren New, San Diego CA, USA (PST)
I ordered stamps from Zazzle that read "Place Stamp Here".
Post a reply to this message
|
|