How do I get the PID of my main process (and not one of its workers)?

Question

I'm using CentOS 7. I'm trying to write a script to start and stop a puma process but I can't figure out how to get the "master" PID, if taht is even the right term. In the below command

[rails@server myproject_production]$ ps aux | grep puma
rails    15767  0.0  1.2 437904 13612 ?        Sl   17:20   0:00 puma 3.11.4 (tcp://0.0.0.0:3000,unix:///home/rails/myproject_production/shared/sockets/puma.sock) [myproject_production]
rails    15779  0.6  7.6 1061248 80688 ?       Sl   17:20   0:05 puma: cluster worker 1: 15767 [myproject_production]
rails    15781  0.6  7.7 1061248 80876 ?       Sl   17:20   0:05 puma: cluster worker 2: 15767 [myproject_production]
rails    15785  0.6  7.4 1061964 78488 ?       Sl   17:20   0:05 puma: cluster worker 3: 15767 [myproject_production]
rails    15880  0.7  7.4 1059612 78592 ?       Sl   17:22   0:05 puma: cluster worker 0: 15767 [myproject_production]
rails    17106  0.0  0.1 112612  1064 pts/0    S+   17:33   0:00 grep --color=auto puma

The master PID is "15767". If I kill that all the other puma processes will die. How do I write a command to get taht into a script variable?

Answer 1

Use pgrep instead of filtering the output from ps. I think in your case, pgrep -f '^([^ ]*/)?puma ' will match the right process, but experiment a bit to make sure that you're getting what you want and no more.

Once you're satisfied that pgrep is finding the process you want to kill, replace pgrep by pkill.

Answer 2

You could do that using ps as follow:

ps -ef | grep puma | grep -v grep | grep -v cluster | awk '{print $2}' | xargs kill

Explaination:
ps: gives the process information
grep: filters on the string provided
grep -v: removes the one with matching records
awk '{print $2}': returns the second item from the line
xargs: converts the input provided into arguments for the command that follows it