Assuming that the
tar process does not terminate when it reaches the end of a tape, but simply emits a request for a new tape, you are going to need to launch tar as a child process of something that is capable of reading tar's standard output and/or stderr stream. The parent process will have to monitor that stream, and test for the presence of the request for a new tape. That process will also have to be able to send the e-mail. For such a task, I would use Perl, but that is simply my weapon of choice; it is probably do-able in bash or any number of other scripting tools.
The meat of the solution would look something like this:
Code:
#! /usr/bin/perl -w
#
# LQzimbot.pl -- launches tar and watches for 'end-of-tape'
#
use strict;
open( TAR, "sudo tar --totals -H pax -cMvf /dev/st0 *|" ) || die "Couldn't launch tar : $!\n";
while( <TAR> ){
if( $_ =~ m/end-of-tape-message/ ){
#
# Your routine to send e-mails here
#
print "Need new tape\n";
}
elsif( $_ =~ /some terminal error message/ ){
print "tar failed\n";
exit(1);
}
}
close TAR;
print "Done\n";
exit(0);
You can fill in the tar messge strings as appropriate. I don't use tapes, so I'm not sure how tar knows if the tape is replaced. If it needs any user input to tell it to carry on, that would complicate matters.
Since all of the hard work is being done in Perl, I would probably write the entire thing in Perl, but you can pass exit status back to the bash script, and embed something like this sample into your shell script instead.
--- rod.
