I have the following Problem, we have a big DB2-Server with 63 Databases.
And the Hardware is crap.

So sometimes the phone rings and someone asks me "Why is the server so slow" and then i had to connect to the DB and start looking.

Now i have the idea to write a script in some language to do this automated for me. So when the phone again rings, i only have to start the script and see why the server is at 100% CPU usage.

I hope someone could help, or give me an idea.

jhp
black_

If you are running DB2 for Linx, Windows, or UNIX then having 63 active databases at one time is not a good idea. Each database uses a certain amount of memory that is not shared between the databases. You might want to consider consolodating databases and using schemas to keep the data seperate.

Its our Development Server so the performance is not so critical.
But if someone locks the server complete up, i want to know who it is.

About the transfer in schemas, i recommend this a time ago, but noone wants it.
And if they are pleased with the performance, why should i keep trying.

jhp
black_

if u running UNIX or linux try this sheel script, then see which process get max CPU tics "C" column



if [ $1 eq '']
then
a=1
else
a=$1
fi

if [ $1 eq '']
then
b=25
else
b=$2
fi

while true
do
vmstat 1 2 > /tmp/process.out
echo "-------------------------------------------------------------------" >> /tmp/process.out
print " ORDER BY CPU Ticks [ C ]" >> /tmp/process.out
print " UID PID PPID C STIME TTY TIME CMD " >> /tmp/process.out
echo "-------- ---- ------ --- -------- ------ ----- --------------------" >> /tmp/process.out
ps -ef | sort -rn +3 | head -n $b >> /tmp/process.out

clear
head -n $b /tmp/process.out
sleep $a
done

Thx for the script, but i forget to mention that the server is UDB 8.1 on Windows 2000 Server.

jhp
black_