![]() | |
#1
| |||
| |||
|
#2
| |||
| |||
|
#3
| |||
| |||
|
#4
| |||
| |||
|
|
We are frequently required to kill the ports for our branch stores who are connected by TCP/IP. We can either shell out to linux, find the process and kill it, often along with a second process which I believe to be the Pick process or I can find the process by typing lu at the pick prompt. There is a "Kill" program in DM which works by typing KILL P23 to kill the linux process for the pick port 23 in showing my age and lack of programming skills, how can I run the program from our application account and secondly, we freqently have to kill the process number, one less than that shown using lu in addition to that process. Would there be a danger in modifying the "KILL" program to kill that previous port as well. It would just be simpler to be able to type KILL P23 from TCL rather than typing sh <return>, finding the port and the related process(s) and then killing them and lastly, typing exit to return to Pick. Thanks for you help or any other suggested remedies. Bob Marik I have a similar problem where I use "nailed" ssh ports for most of my |
#5
| |||
| |||
|
|
rmarik wrote: I have a similar problem where I use "nailed" ssh ports for most of my users. If for any reason they get knocked of without exiting, I have to kill the Linux process for the port they're on. I keep wondering if I shouldn't go through D3 for a more controlled exit. I "nail" the ssh ports by putting a script in their profile, when they login, the script connects them to a specific D3 port. If the port is unavailable they get bounced right out. |
#6
| |||
| |||
|
#7
| |||
| |||
|
|
rmarik wrote: We are frequently required to kill the ports for our branch stores who are connected by TCP/IP. We can either shell out to linux, find the process and kill it, often along with a second process which I believe to be the Pick process or I can find the process by typing lu at the pick prompt. There is a "Kill" program in DM which works by typing KILL P23 to kill the linux process for the pick port 23 in showing my age and lack of programming skills, how can I run the program from our application account and secondly, we freqently have to kill the process number, one less than that shown using lu in addition to that process. Would there be a danger in modifying the "KILL" program to kill that previous port as well. It would just be simpler to be able to type KILL P23 from TCL rather than typing sh <return>, finding the port and the related process(s) and then killing them and lastly, typing exit to return to Pick. Thanks for you help or any other suggested remedies. Bob Marik I have a similar problem where I use "nailed" ssh ports for most of my users. If for any reason they get knocked of without exiting, I have to kill the Linux process for the port they're on. I keep wondering if I shouldn't go through D3 for a more controlled exit. I "nail" the ssh ports by putting a script in their profile, when they login, the script connects them to a specific D3 port. If the port is unavailable they get bounced right out. |
#8
| |||
| |||
|
| Originally Posted by Ross Ferris from tcl in your data account: copy dm,, kill to mdjob done! |
#9
| |||
| |||
|
|
Bruce, have you tried the auto-xinet script in dm,bp,? You need to run it from a d3 session started from root with the "-l" option, and you will need to modify it for ssh instead of telnet (is ssh really necessary for local users?). I used auto-xinet (with telnet) to set up nailed ports at several sites with minimal problems (also I put 'dcd-on' in the logon procs (and usually also start d3 with the -dcdon option), and 'trap dcd exit' in user-coldstart). There are the occasional problems with the d3 session not disconnecting, but infrequent enough that it isn't a big problem; I show someone on site (like the office manager?) how to use my "fix.port" tool: * clear hung telnet port * 07-12-02 asb tclread buf buf = trim(buf) port = field(buf," ",2) if port eq "" or not(port matches "0n") then print "Enter port number to fix: ": prompt "" input port call input.test(port) if port eq "" or port eq "quit" then stop if port matches "0n" else print "Port number must be a NUMBER" stop end end if port eq system(22) then print "You cannot fix.port on yourself!" stop end cmd = "reset-user " ort**print cmd execute cmd cmd = "kill p" ort**print cmd execute cmd capturing xxx call get.user.srt(user) date.time = date():".":time() open "asb,fix.ports," to fix.ports then rec = port rec<2> = user write rec on fix.ports,date.time end /Scott Ballinger Pareto Corporation Edmonds WA USA 206 713 6006 bruce ackman wrote: rmarik wrote: I have a similar problem where I use "nailed" ssh ports for most of my users. If for any reason they get knocked of without exiting, I have to kill the Linux process for the port they're on. I keep wondering if I shouldn't go through D3 for a more controlled exit. I "nail" the ssh ports by putting a script in their profile, when they login, the script connects them to a specific D3 port. If the port is unavailable they get bounced right out |
![]() |
| Thread Tools | |
| Display Modes | |
| |