Linux: ps command's output is randomly different than expected, can you find the reason?

Question

I wrote a Nagios check which counts how many pgpool processes are running and how long each process runs. I'm running on Linux CentOS 6.4

In order to get the time the process is running, I'm using the next command:

ps -p PID -o etime=

Example:

[root@pgpool ~]# ps -p 28737 -o etime=
   08:35:48
[root@pgpool ~]#

This is the normal output, but sometimes I get the following output and it break the script: 7-17:15:52 Example:

[root@pgpool ~]# for prc in $(ps -ef | grep pgpool | grep -v wait | grep -v PCP | awk '{print $2}'); do ps -p $prc -o etime= ; done
      40:55
   22:08:43
      23:55
   15:12:36
      20:35
 7-17:15:52
   09:34:35
29-00:56:18
[root@pgpool ~]#

So my questions are:

What does it mean? that the process is running between 7 to 17 hours?
How come the output layout is different at some times?

You have your answer. What you didn't ask was how to do this better. Use etimes to get seconds. Then you have no weird parsing of the time to do. — Zan Lynx, Jan 05 '15 at 19:36
I've tried `ps -p PID -o etimes` and `ps -p PID -o etimes=` but none of them worked... — Itai Ganot, Feb 24 '15 at 16:38
It works for me. Tested on Ubuntu 14 LTS and Fedora 21. Maybe CentOS 6 is just too old. I didn't realize it was a new thing. — Zan Lynx, Feb 24 '15 at 16:56

score 13 · Accepted Answer · answered Jan 05 '15 at 08:33

13

man ps, section etime:

etime ELAPSED elapsed time since the process was started, in the form [[dd-]hh:]mm:ss.

So your process runs for longer than a day.

answered Jan 05 '15 at 08:33

Sven

98,649
14
180
226

Linux: ps command's output is randomly different than expected, can you find the reason?

1 Answers1