accurately timing an execution of a system call using perl
Is there a better way programmatically using perl to track the timing of a unix process (start time, end time)??
For example, I have a perl script that launches a unix command using system(). I print the timestamp before the system() and after.
My log of the start time and end time are the same. I know for sure that when the process completes its a matter of minutes, so its not accurate.
Here's my code thus far:
my $dt = strftime('%Y%m%d', localtime(time));
my $dt2 = strftime('%Y%m%d-%H:%M:%S', localtime(time));
my @tbls = qx(mysql -u foo -pmypw --database $dbsrc -h $node -ss -e "show tables");
open (CRLOG, ">dump-$dt.log") || die "cannot append";
foreach my $tbls (@tbls)
{
chomp $tbls;
print CRLOG "TIME START => $dt2\n";
my $crfile = qq{mysql -u foo -pmypw --database $dbsrc -h $node.test.com -ss -e "SELECT 'a','b','c' UNION SELECT 1,2,3 FROM $tbls"| tr "\t" ",">$node-$tbls.$dt.csv}; system($crfile);
开发者_StackOverflow中文版 print CRLOG "COMMAND => $crfile\n";
print CRLOG "TIME END => $dt2\n";
};
close (CRLOG);
$dt2
is a constant, but it looks like you want it to recompute the timestamp every time it is used. For that purpose, you should use a function.
sub dt2 {
return strftime('%Y%m%d-%H:%M:%S', localtime(time))
}
...
open (CRLOG, ">dump-$dt.log") || die "cannot append";
foreach my $tbls (@tbls)
{
chomp $tbls;
print CRLOG "TIME START => ", &dt2, "\n";
my $crfile = qq{mysql -u foo -pmypw ...};
system($crfile);
print CRLOG "COMMAND => $crfile\n";
print CRLOG "TIME END => ", &dt2, "\n";
};
close (CRLOG);
Have you tried using the "time" command to time the command execution?
http://www.computerhope.com/unix/utime.htm
So in your perl script it would be something like this to capture the output:
$time_me = `/usr/bin/time $my_command`;
Then parse the $time_me variable to get the values.
精彩评论