It is required to trap the signal send to a daemon process before rebooting a unix server. Suppose a script abc.ksh is running in the server as daemon. Before rebooting the server, the unix admin kills all the daemon processes. It is not known to me how admin kills the processes; I mean by which kill option (e.g. -9). And if it is killed by -9 option, what will be the trap signal in the script. I tried with 1, 2, 9, 15. But it is not working.

Thanks in advance.

You cannot trap signal 9. You could have a watchdog process running for the server. When the watchdog process detects the daemon is no more, you can do the necessary things.

Thanks vino. But the problem with the watchdog script is it might get killed under similar circumstances; I mean when a unix server is preparing for rebooting. So, it doesn't seem useful solution.

What you are describing is SUPPOSED to happen.

If it is imperative that the abc.ksh daemon do something special before exiting, then you will have to have the sysadmin run a special script for shutdown, one that calls your abc.ksh, then waits 1 - 2 seconds then shutesdown the system. Otherwise why are you worried about it?

The constraint is I can not ask sysadmin to do something special for my daemon script abc.ksh. Let me tell you in detail. The abc.ksh is a monitoring script which checks whether database is up and running. It sleeps for 10 mins and wakes up to do these tests and again sleeps. If database is not running then it sends mail to related persons to look into the matter.
During last weekend the unix server is rebooted without my knowledge. Yesterday only I discovered the daemon is not running. So, I need to have a trap in my script which will recognize it is being killed. And subsequently it will send a mail to me informing me that the daemon is not running. I am facing problem with the trap signal. I have given 1, 2, 3, 6, 9, 15, 18, 20, 32 and tested with control C. The script gets killed and sends me a mail. But when I am testing it with kill -9, no mail is being sent.

I want to know --
1. Which option sysadmin use when he/she kills unix daemons before rebooting?
2. What will be trap signal if it is being killed by kill -9 option.

As said before, you cannot trap SIGKILL or SIGSTOP by anyway,

I can suggest a different idea,

since the main problem is if the server gets rebooted without notification there is no way to identify that the daemon has to be specifically rebooted.

Instead have your daemon in rc scripts when server goes to a reboot from shutdown automatically daemon also will start running.

Hope this helps!

Why dont you put the daemon script in the cron job ? In that case, you dont have to worry about the machine restarts. Also, you can avoid the sleeping state of the script.