linux - ps start_time as timestamp -

i'm using ps command monitor processes. however, like have time output unix timestamp.

the command using:

ps -eo class,s,user,pid,pcpu,stat,start_time,cputime,args --sort=class | \   awk '$1 != "rr" && $1 != "ff" {print $0; next } 0' 

thanks!

use etimes option (elapsed time since process started, in seconds) , awk's systime() function returns current timestamp in seconds.

suppose 7th column etimes, awk expression should this:

nr == 1 ? "timestamp" : systime() - $7 

example:

ps -eo class,s,user,pid,pcpu,stat,etimes,cputime,args --sort=class | awk \   '$1 != "rr" && $1 != "ff" {print $0, nr == 1 ? "timestamp" : systime() - $7 }' 

sample output

cls s user       pid %cpu stat elapsed     time command timestamp ts  s root         1  0.0 ss      1717 00:00:01 init [3] 1479001705 ts  s root         2  0.0 s       1717 00:00:00 [kthreadd] 1479001705 ts  s root         3  0.0 s       1717 00:00:00 [ksoftirqd/0] 1479001705 

if want replace seventh column, can override $7 variable: $7 = systime() - $7.

if result of command supposed parsed further, remove headers --no-headers option, e.g.:

ps --no-headers -eo class,s,user,pid,pcpu,stat,etimes,cputime,args \   --sort=class | awk '$1 != "rr" && $1 != "ff" { $7 = systime() - $7; print }' 

sample output

ts s root 1 0.0 ss 1479001705 00:00:01 init [3] ts s root 2 0.0 s 1479001705 00:00:00 [kthreadd] ts s root 3 0.0 s 1479001705 00:00:00 [ksoftirqd/0] 

note, there no need next statement in script, since purpose stop processing current record , go on next record, , there no statements (pattern-action pairs) skip @ end of script.