Unable to use AWK redirection

Question

I want to use the AWK redirection feature and what I've done so far is this :

$ vmstat 1 | awk ' { print $2 > "outfile" } '

*Actually the commands before awk are a lot more complicated , but it's a simplified demonstration.

If I run the above command without redirection , I would get the desired result in the stdout. But after redirecting it to outfile , it's still empty :

$ cat outfile
$

What's wrong with that ?

TIA.

Answer 1

awk buffers its output. If your awk implementation provides it (as gawk, mawk¹, nawk, and BSD awk do), use fflush().

  fflush([file])        Flush any buffers associated with the open output file 
                        or pipe file.  If file is missing or if it is  the null 
                        string,  then  flush  all open output files and pipes.

So, write it this way:

vmstat 1 | awk '{print $2 > "outfile"; fflush()}'

The GNU awk manual I/O section on fflush is a worth reading. There you will also find that fflush has been accepted to the next POSIX standard.

---

As an extra, notice that you can give the number of samples that vmstat should output. So, if you want only 5 samples (for example), you can wait the 5 seconds until the command terminates and then the file will contain the output:

vmstat 1 5 | awk '{print $2 > "outfile"}'

---

¹With mawk the syntax is a bit different: mawk -W interactive '{print $2 > "outfile"; fflush("")}'.

Answer 2

The problem is buffering, it can be disabled.

vmstat 1 | stdbuf -o0 awk '{print $2}' >> out

• -o output stream
• 0 stream will be unbuffered

or you can call vmstat in while loop with sleep 1

while true; do 
     vmstat | awk '{print $2}' >> output.file
     sleep 1
done