Shell Programming and Scripting

Hello all,

This is my requirement:

1. I have 6 VMs running Apache (for Oracle EBS) as Linux user oracle.

2. From a central server (VM), I need to login to all the 6 VMs as oracle user (I have already set up ssh equivalence, so it is password less authentication).

3. Find the number of processes running as oracle user on each VM and if less than 20, something is wrong and the same information is written to a log file on the central server.

4. If it is more than 20 processes, then everything is fine.

5. My question or trouble is: How do I know that collectively, Apache is running fine for this application (Oracle EBS)?


TIA,

Regards,

Praveen

You could try something like this:

Code:

ERR=0
for HOST in server1 server2 server3 server4 server5 server6
do
    NUMPS=$(ssh oracle@$HOST ps -fu oracle | wc -l)
    if [ $NUMPS -lt 20 ]
    then
        echo "Error with oracle on $HOST ($NUMPS vs 20)"
        ERR=1
    fi
done

if [ $ERR -eq 0 ]
then
    echo "Collectively, Apache is running fine"
fi

Code:

ps -u oracle | wc -l

is less overhead than

Code:

ps -fu oracle | wc -l

When running as oracle, it eventually might count itself.

Thanks for the replies.

My understanding is that any variable that is instantiated inside a while loop can not be used outside the loop.

So, the workaround would be to redirect the variable value to a file. Read the file outside the while loop, assign the value stored in the file to a variable and then compare it with the reference.

I feel this is a bit of a crude way to do it but it should work.

Is there any other way that it can be done?


Regards,

Praveen

True for a while loop. It runs in a sub shell, i.e. works with a copy of the variable. A for loop does not run in a sub shell.
I recommend to install xymon, and use the xymon agents for Oracle (run as user xymon).
Xymon has got a web interface: global status, drill-down by a simple click. Also: alarm history, availability reports, trend analysis.