In-depth understanding of the meaning of 2>&1 in the Linux shell (the most complete network, you can understand after reading it)

What do 1 and 2 stand for in Linux

In Linux system 0 1 2 is a file descriptor

It can be seen from the above table that we usually use

echo "hello" > t.log 

In fact, it can also be written as

echo "hello" 1> t.log

About the meaning of 2>&1

I won’t elaborate on input/output redirection in this article. If you don’t understand, please refer to here: Shell: Pipes and Redirection

• Meaning: Redirect standard error output to standard output

• The symbol >& is a whole and cannot be separated. After being separated, it will not have the above meaning.

  • For example, some people might think like this: 2 is standard error input, 1 is standard output, and> is a redirection symbol. Then, should "redirect standard error output to standard output" be written as "2>1"? Is that right?
  • If you have tried it, you will know that the wording of 2>1 actually redirects the standard error output to a file named "1".
• It’s not possible to write 2&>1

Why 2>&1 should be put at the back

Consider the following shell command

nohup java -jar app.jar >log 2>&1 &

(The last & means put the command in the background for execution, it is not the focus of this article, you can Google it yourself if you don’t understand it)

Why 2>&1 must be written after >log, which means that both standard error output and standard output are directed to log?

We might as well understand 1 and 2 as a pointer, and then look at the above statement is like this:

本来1----->屏幕 （1指向屏幕）
执行>log后， 1----->log (1指向log)
执行2>&1后， 2----->1 (2指向1，而1指向log,因此2也指向了log)
``
再来分析下

nohup java -jar app.jar 2>&1 >log &

本来1----->屏幕 （1指向屏幕）
执行2>&1后， 2----->1 (2指向1，而1指向屏幕,因此2也指向了屏幕)
执行>log后， 1----->log (1指向log，2还是指向屏幕)

So this is not the result we want.

Simply do a test to test the above ideas:

The java code is as follows:

public class Htest {
    public static void main(String[] args) {
        System.out.println("out1");
        System.err.println("error1");
    }
}

After javac is compiled, run the following command:

java Htest 2>&1 > log

You will see only "error1" is output on the terminal, and only "out1" in the log file

It is too troublesome to write ">log 2>&1" every time, can you abbreviate it?

There are two abbreviations

&>log
>&log

For example, the wording in the above section can be abbreviated as:

nohup java -jar app.jar &>log &

The above two methods have the same semantics as ">log 2>&1".

So is there a difference between &> and >& in the above two methods?

There is no difference in semantics, but the first method is the best choice, and the first method is generally used.

Author: A Walking People
Original: https://blog.csdn.net/zhaominpro/article/details/82630528