According to my SuSE Linux 8.2 Administration Guide, the logrotate program is provided.

You don't necessarily need to use the system file /etc/logrotate.conf. You can instead use your own local config file, specific for your application, and just specify that config file on the command line. Personally, I would write a shell script (sorry, no perl) that runs from cron, which invokes logrotate and then does the copying you want. Run it just before the bosses get to work in the morning :-)

From Mark Jason Dominus' Lightweight Databases talk:

tie @LOG, 'Tie::File', '/etc/logfile';

sub log {
    push @LOG, @_;
    my $overflow = @LOG - 100;
    splice @LOG, 0, $overflow if $overflow > 0;
}
[download]

His solution involves tie-ing the log file to an array using Tie::File, which is a core module.

That doesn't work if the file is continuously being updated by another process (or processes), which is often the case with logfiles. It's only a solution if you are in full control over the single process that's updating the log file. Even if there are two instances of the same program log to the file, the given solution is going to fail.

I think i give it a try afaik the "logfile" is produced at a given time once a day only so if its possible that the file isnt altered at that few seconds it is ok ... I think it might be a cron job, but i am searching for that ...

Thanks in Advance MH

I don't know much about concurrent access to files, but wouldn't locking the file solve this issue?
Blog

Sure, it would be safer and cleaner with modules, but without modules is still easier than remote administering from a windows box (remember you might have to adapt some of the modules to do things remote, which would invalidate that "safe and clean").