Bash: extract data from a command-line program

Question

I know how to get the total memory in the system:

$ free -lm
             total       used       free     shared    
Mem:          3008       2495        513         57

I know how to get the main memory consumption (RSS) of worker processes in Nginx:

$ ps -C nginx -O rss
  PID   RSS S TTY          TIME COMMAND
 1564  1336 S ?        00:00:00 nginx: master process /usr/sbin/nginx
 1565  1756 S ?        00:00:00 nginx: worker process
 1566  1756 S ?        00:00:00 nginx: worker process
 1567  1756 S ?        00:00:00 nginx: worker process
 1568  1756 S ?        00:00:00 nginx: worker process

Now to determine how many worker processes my system could use without resorting to swap:

echo $((3008 * 1024))
3080192
$ echo $((3080192 / 1756))
1754

My server can handle 1754 nginx workers without resorting to swap. However, it would be great if I can take this multi-step process above and make it a single line which can be executed from command line.

My problem is I don't know how to extract that "3008" from the free command from the command line. How can I tackle this?

Answer 1

My problem is I don't know how to extract that "3008" from the free command from the command line

Given this output:

free -lm
             total       used       free     shared    buffers     cached
Mem:          3757       1765       1991        138        122        766
Low:          3757       1765       1991
High:            0          0          0
-/+ buffers/cache:        876       2880
Swap:         7772          0       7772

Try this:

free -lm | grep '^Mem' | awk '{ print $2 }'
3757

Which will return the total column in the Mem: row. In my case 3757.

Answer 2

Piping free to awk will extract the value you require:

free -l | awk '/^Mem/{print $2}'

Answer 3

Something like:

echo $(($(($(free -lm | grep Mem | awk '{print $2}') * 1024)) / $(ps -C nginx -O rss |grep 'nginx: worker process$' | awk '{print $2}' | tail -1)))

Answer 4

You should be able to do this with head, tail and cut:

free -lm | head -2 | tail -1 | tr -s ' ' | cut -f2 -d' '

head -2 because it is the top two lines.

             total       used       free     shared    buffers     cached
Mem:          3757       1765       1991        138        122        766

tail -1 because it is the last line of that.

Mem:          3757       1765       1991        138        122        766

tr -s ' ' - converts chains of whitespace to a single space.

Mem: 3757 1765 1991 138 122 766

Finally - cut -f2 -d' ' - getting the second field, space delimited.

3757