I am at a loss on how to even start this Perl script. I have the following processes:

Command used to get this info:

ps aux --forest | grep -e process_name -e ksh | awk '{if ($1 == "user1" && $1 != "root" && $1 != "UID" && $1 != "xfs" && $1 != "mfg" && $1 != "mfgnet") print $0}'

Processes
user1 2819 0.0 0.0 4272 612 ? S Nov17 0:00 \_ -pksh-ksh
user1 2820 0.0 0.0 64956 1584 pts/833 Ss+ Nov17 0:00 \_ -ksh

user1 2918 0.0 0.0 4272 616 ? S Nov17 0:00 \_ -pksh-ksh
user1 2922 0.0 0.0 64956 1620 pts/689 Ss+ Nov17 0:00 \_ -ksh
user1 6657 0.0 0.4 1322176 140728 pts/689 S+ Nov17 0:00 \_ /usr/bin/process_name

I need a way to kill off the pids 2819, 2820 because they do not have a process tied to them like pids 2918, 2922 and 6657.

The way it works is peek shell (pid 2918)is opened then it starts a ksh (pid 2922) session then from there the end user runs a command (pid 6657).

Is there a way to do this. All I need is an idea to get started the rest I can do.

Actually, the way we usually do things is you get started and run into a difficulty, and post your code here, and we help out with specific questions.

But if you want to get started, do this:
... and go from there.

@wje_lq - haha very funny!

I know that is protocol. I am having a brain cramp on how to started to basically seperate the pids with 2 lines (pksh, ksh process) and the pids with 3 lines (pksh, ksh, process_name).

Any real help will be appreciated.

So wait. Your input data comes in paragraphs of either two or three lines, followed by a blank line. In the paragraphs consisting of exactly two lines, you want to kill the processes numbered at the beginning of each line. You want to drop on the floor the paragraphs consisting of three lines. You want to puke if a paragraph consists of fewer than two lines or more than three lines.

Is this correct?

This is where my data comes from.

It returns this output. Notice I put a break between the ones that have a pksh,ksh,process and the only one that does not have all three. The very last one (in bold) that only has pksh, ksh those are the ones I need to kill off via a script.

And I need to do this for hundreds of users not just one user, thus the script.

To the OP: do it the Perl way: http://search.cpan.org/~durist/Proc-...rocessTable.pm . I.e. you do not need 'ps'. 'grep', 'awk' - everything can be done from inside Perl.

2 members found this post helpful.

Thanks! That is the type of answer I was looking for.

Ok I have time to look at this again. I have created some perl code that gives me this:

user1 13794 13493 2010 ? 09:09:45 2011 pts/3874 -pksh-ksh
user1 13795 13794 2010 ? 09:09:45 2011 pts/3868 -ksh
user1 13854 13795 2010 ? 09:09:45 2011 pts/3868 db_process

user1 10094 10093 2010 ? 09:09:45 2011 pts/3874 -pksh-ksh
user1 10095 10094 2010 ? 09:09:45 2011 pts/3868 -ksh

I thought the "?" meant it was dead but it does not because I can connect to the 13794 Peek session and it has the question mark.

I need to be able to kill the second group because it has no database connection and from what we can tell it is dead. If we try and reconnect to that session id via Peek it states this session is no longer valid. The one with the db_process is able to be connected to. This would not be such an issue but we only have 1600 Peek connections and need to keep this tidy. If this was killing one process or a few I could do it by hand but it is 100's.

So how do I get my perl code to check this line:
user1 13794 13493 2010 ? 09:09:45 2011 pts/3874 -pksh-ksh

and grab the 13794 id and then check line number two in the third column for this very same number 13794 and then check the 2nd column, 2nd line id 13795 and see if it matches column 3, line3 13795.

I was told to use Proc::ProcessTable but I am not sure how to kill off the ones I want with this output.

This is above my scripting know how and it would be great if someone out there could give me some direction.

Thanks in advance!