Linux命令（cp，rm）不在perl脚本中执行。 一些作品。 但是没有错误返回

Question

I'm having a very strange error.

I run a perl script which executes linux commands. They are executed like this:

my $err = `cp -r $HTML /tssobe/www/tstweb/$subpath/$HTMLDIR1`;
myLog("$err");

And $err is empty, which mean the command didn't return and error. (right?)

I tried to execute the linux command with exec "" or system (), but no success. I tried to change the path. Same.

Also, I tried to run only the cp command in a new perl script. It works. But not in my full perl script.

In this perl script, some commands are working, some are not.

The script was working yesterday, Not anymore this morning. No changes have been made in the meantime.

I tried a lot of things, I would be glad if anybody has an idea.

EDIT: The server was having a lot of processes unterminated. Cleaning those solved the problem. So the problem is related to another application, but I'll improve the logging thanks to your comments.

Answer 1

Small problem: you are NOT capturing STDERR, so you won't see the error (you are also not checking $? return code).

You should do

my $err = `cp -r $HTML /tssobe/www/tstweb/$subpath/$HTMLDIR1 2>&1`;

to redirect STDERR to STDOUT, or use one of the modules for running commands.

---

Large problem:

You should not run system commands from Perl for which Perl-native modules exist. In this case: File::Copy::Recursive module.

You can also roll your own directory copied from File::Copy .

Answer 2

Are you using backticks? Add -v to the cp commmand to see something in STDOUT and redirect the STDERR to STDOUT and check the cmd exitcode not the error message in the STDERR.

What about printing out the command output right after the execution?

my $err = `cp -rv $HTML /tssobe/www/tstweb/$subpath/$HTMLDIR1 2>&1`;
my $exitcode = $? >> 8;
warn "Output: $err\nexitcode: $exitcode\n";

It would be better to use qx. Check this: http://www.perlmonks.org/?node_id=454715

Answer 3

You may also want to quote arguments that may potentially contain any shell special characters, including spaces. As shell will do word splitting on the string given to it, if $HTML contains a space cp would get more arguments than you expect. Perl has very simple mechanism for that: \\Q\\E . Here is how you do it:

my $err = `cp -r \Q$HTML\E \Q/tssobe/www/tstweb/$subpath/$HTMLDIR1\E 2>&1`;

Anything except for alphanumeric will be backslash escaped before passing it to the shell. And you would provide exactly 2 arguments to cp regardless of what is in those variables.