I have a plugin in nagios "check_icmp" who return 4 values if the command can ping the demanded host however i got only 1 values if the ping failed so i used the following command :
'/usr/lib64/nagios/plugins/check_icmp -w 1000.0,20% -c 1400.0,60% -H 8.8.4.5 -m 5 | sed 's/pl=100%;20;60;0;100/rta=nan;;;; rtmax=nan;;;; rtmin=nan;;;; pl=100%;20;60;0;100/g'
and it returns
CRITICAL - 8.8.4.5: rta nan, lost 100%|rta=nan;;;; rtmax=nan;;;; rtmin=nan;;;; pl=100%;20;60;0;100
instead of
CRITICAL - 8.8.4.5: rta nan, lost 100%|pl=100%;20;60;0;100
so it works great on the host but if i put this command in nagiosql the current status stay in green "OK" even if the ping failed :
I would guess that the reason that you're getting an OK despite the fact that the result is "CRITICAL" is because the only thing Nagios cares about is the return code of the command you ran (also called an exit code). I would guess that sed has the last word, exits with return code 0 which means "everything went fine", and it did obviously: for sed.
Since you don't explain why you're doing any of this, I can't comment on whether this is a good idea or not, but it seems very odd in my mind to include a pipe to sed in your check command. I understand that you are sanitizing the performance data that is returned by the plugin, but why?
Nagios is built to interpret return codes as the status of your check, that's just the way it is. Whatever issues you are seeing because of this plugins behavior specifically surely can be resolved in another way without running sed magic on your checks that you may or may not get intended results from. It's also worth noting that if you're inserting this into RRD files, messing with the order or amount of values may create headaches, so I really wouldn't recommend it.
As a constructive suggestion in the future, please include what brought you to your current solution when asking these questions, as you can never know whether it's actually the best way to resolve your original issue. This is often called an XY Problem.