Does anyone know from least harsh to most harsh which kill signals are easiest and hardest on the system? The reason I ask is that I think we issue too many "kill -9" commands. I theorize that by doing that we create either a memory leak or never releases DB2 connections that eventually takes our system to its knees. I try to kill processes with a -2 usually, but now I believe a -15 was more in order. My boss & other co-workers are a little trigger happy, and use -9.

Does this sound most reasonable?
-- start with -15
-- wait a few seconds, recheck running processes
-- try -15 again
-- wait...check
-- if it's not down issue a -2
-- wait...check
-- -2 again
-- wait...check
-- THEN we may start using -9......or not? Ideas?

me@mylinuxbox:/home/me> kill -l
1 HUP Hangup
2 INT Interrupt
3 QUIT Quit
4 ILL Illegal instruction
5 TRAP Trace trap
6 ABRT Abort
7 BUS Bus error
8 FPE Floating point exception
9 KILL Killed
10 USR1 User defined signal 1
11 SEGV Memory fault
12 USR2 User defined signal 2
13 PIPE Broken pipe
14 ALRM Alarm clock
15 TERM Terminated
16 STKFLT Stack fault
17 CHLD Child exited
18 CONT Continued
19 STOP Stopped (signal)
20 TSTP Stopped
21 TTIN Stopped (tty input)
22 TTOU Stopped (tty output)
23 URG Urgent I/O condition
24 XCPU CPU time limit exceeded
25 XFSZ File size limit exceeded
26 VTALRM Virtual timer expired
27 PROF Profiling timer expired
28 WINCH Window size change
29 IO I/O possible
30 PWR Power-fail/Restart
31 UNUSED Unused
32 32 Signal 32
..then 33 thru 63

You approach is very reasonable to me, people who directly start with -9 are fools as -9 is a last resort feature, not a general purpose one.

By the way, you may want to use the non numerical names for better portability, i.e. kill -TERM, kill -INT then kill -KILL.