store os-command output to a variable

[dlangschied]

I do not think that this is possible, 'cus Lord knows I have tried, but I will throw it out there.

I would like to store the output of an os-command into a variable. I know that I could have the output stored to a file on the os and the content read back into the procedure, but I thought that maybe I could make it cleaner.

Situation:
I have a user that has a session that needs to be killed. I can only kill in promon as root, but that is dangerous to let the support group do that.
I have a procedure that, when run, generates a list of users to disconnect. The support group selects the user and a cron job, run as root, performs the kill in promon using a keyfile generated by the procedure.

Problem:
Recently we addded ODBC to the mix and now we have a single db user and all others show up in promon as blank (curses!).

Proposed resolution:
Go out and match the pid with the user that created it on the OS. This will now allow the list to contain the info that the support group needs; problem solved.

So...
I know what I need to do. I just want to make it as clean as possible. If I get the pid from the os, using os-command, can I store it directly to a variable (or am I SOL?)

Any suggestions?

 

[MaximMonin]

Linux:
OS-COMMAND VALUE ("sudo sh -c ""proshut dbname -C disconnect " + string (id-user) + """").

Make sure change /etc/sudoers file

#This file gives a right to any user of group "progress" to execute sudo command and disconnect process.
# User privilege specification
root ALL=(ALL) ALL
progress ALL=(ALL) ALL


# Uncomment to allow people in group wheel to run all commands
# %wheel ALL=(ALL) ALL
progress ALL=(ALL) NOPASSWD: ALL

 

[dlangschied]

This info, though informative, misses the mark. My issue is that I do not no the User's ID (that being the OS id of the user logging into the application). I need this information for support to identify the progress user-id to select for dicsonnect. And that is what I am looking for.

Here is a bit of the code that I have started with.

[FONT=r_ansi][FONT=r_ansi]def var get-pid-user as char format "x(70)".
assign get-pid-user = "ps -ef | grep _progres | grep ".
[FONT=r_ansi][FONT=r_ansi]For Each _Connect WHERE _connect-Type = "SELF" no-lock:
if _Connect-Name = " " then do:
[/FONT][/FONT]assign get-pid-user = get-pid-user + " 13913".
assign get-pid-user = get-pid-user + " | awk -F~" ~" ~'~{print ~$1~}~'".

os-command silent value(get-pid-user).
[/FONT][/FONT]

 

[dlangschied]

Maxim,
The issue is not how to disconnect a user. It is that the user name does not exist. Our support group is not going to know who to disconnect. This is why I have to first sync the _Connect.PID value to the user on the OS to which that PID belongs. Here is a snippet of the code:

[FONT=r_ansi][FONT=r_ansi]def var c-user as char.
def var c-user1 as char.
def var c-user2 as char.

c-user1 = "ps -ef | grep ".
c-user2 = " | grep _progres | awk -F~" ~" ~'~{print $1~}~'".

For Each _Connect WHERE _connect-Type = "SELF" no-lock:
if _Connect-Name = " " then do:
c-user = c-user1 + string(_Connect-Pid) + c-user2.
/*Now somehow assign "os-command silent value( c-user)" output to a variable so that I can pass the user id to a
selectable list for support. Then run proshut with disconnect option for the user's progress id*/
end.
end.
[/FONT][/FONT]

 

[GregTomkins]

You know about INPUT THROUGH x, right, where 'x' can be any Unix command, and you can then do IMPORTs or READKEYs to process the response from Unix?

 

[MaximMonin]

field _Connect._Connect-Usr is user-id to disconnect.

define variable user-id as integer.
For Each _Connect WHERE _connect-Type = "SELF" and Connect-Name = " " and _Connect-PID = process-pid no-lock:
user-id = _Connect._Connect-Usr.

OUTPUT THROUGH value(yourappname with root right).
put unformatted user-id.
OUTPUT close.
end.

but if you just plan to read output then:
define var outputvar as character.
input through value (c-user).
import outputvar.
input close.

 

[dlangschied]

Greg,
This is what I am looking for! I am not familiar with the INPUT FROM command. I have not yet tried it, but if it will allow me to run my OS command and collect the output from that command, super!

Thanks!