On Feb 12, 2008 1:08 PM, Kashif Salman <[EMAIL PROTECTED]> wrote:
>
> On Feb 12, 2008 9:58 AM, Chas. Owens <[EMAIL PROTECTED]> wrote:
> > On Feb 12, 2008 12:38 PM, Michael Barnes <[EMAIL PROTECTED]> wrote:
> > snip
> > > I'm the new kid and this is a beginners forum, so I welcome all ideas
> > > and options. Forgiving my ignorance, would you mind giving an example
> > > of how I would do this with lsof?
> > snip
> >
> > This only works on systems with lsof. Checking file size change is a
> > fairly portable, if inexact, way of checking to see if a file is still
> > being written to. Of course, lsof also has problems: what if the file
> > is opened, written to, and closed, opened, written to, closed, etc.
> > and the lsof runs while the file is closed?
> >
> > #!/usr/bin/perl
> >
> > use warnings;
> > use strict;
> >
> > my $lsof = "/usr/sbin/lsof";
> > my $file = "/tmp/foo.$$";
> > my $pid = fork;
> > die "could not fork" unless defined $pid;
> >
> > unless ($pid) {
> > #child writes to the file for 5 seconds
> > open my $fh, ">", $file
> > or die "could not open $file\n";
> > for my $i (1 .. 5) {
> > print $fh "$i\n";
> > sleep 1;
> > }
> > exit;
> > }
> >
> > #parent monitors file
> > my $file_holder_pid = qx($lsof -t $file);
> > chomp($file_holder_pid);
> > while ($file_holder_pid) {
> > print "$file is current being held open by $file_holder_pid\n";
> > sleep 1;
> > $file_holder_pid = qx($lsof -t $file);
> > chomp($file_holder_pid);
> > }
> >
>
> Yes, it does make the script tied down to mostly UNIX/Linux systems. I
> assumed once the file is dumped in, it won't be accessed by that same
> process so no write close, write close operations on the file.
>
Yeah, the lsof solution isn't bad, it just isn't the right tool all of
the time, and neither is the size monitoring solution. They both have
their own strengths and weaknesses and it is important to know what
they are. There is a third option that solves the problems that both
of the previous solutions have: a signal file. You can be sure the
file is finished being written if you can arrange for the upstream
process to write a zero byte file when it is finished writing. Of
course, it has the drawback that you need to be able to change the
upstream code to send the signal file (which is often impractical).
--
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
http://learn.perl.org/
