Re: Beagle changes ctime

[Joe Shaw <joe ximian com>]

Hi,

On Wed, 2007-03-14 at 22:06 +0100, Felix E. Klee wrote:
> By how much, approximately?  Can I benchmark this?

I don't know about exact numbers, but you could do us a huge service by
measuring it yourself.

Set up a decently sized sandbox somewhere (where the content of the data
is unlikely to change).  I usually use something like /tmp/sandbox and
copy a bunch of data in it.  Make sure that wherever it is has xattr
support on in the filesystem.

Run:

        BEAGLE_EXERCISE_THE_DOG=1 BEAGLE_HOME=/tmp/sandbox beagled --fg --debug --indexing-test-mode --backend files
        
and see how long it takes to run.  Do it a few times,
removing /tmp/sandbox/.beagle before each run.  Then run it a few more
times with BEAGLE_DISABLE_XATTR=1; contrast the running times.  That'll
give you (and us!) a good idea of what the difference in time is.

As Pat said, things will be somewhat slower, but not unusably slow.

> Is it better then to run beagled not in the background and only, say,
> once a week or once a night with "BEAGLE_EXERCISE_THE_DOG=1"?

I don't think you'll need to make any changes to how you run it now.

> I'll also double check the backup tool that I'm using: It doesn't make
> sense to back up an entire file if only its ctime has changed since the
> last incremental backup.

True, although having the accounting information (owner, mode, ACLs,
etc.) is useful information to have backed up, and I can understand why
ctime is used here.

Joe