I'm running a cluster of RH7.1 servers and am doing a significant amount of network socket testing--running network load generators against SMTP, POP, HTTPD, WHOIS servers--until they scream for mercy.
Not surprisingly, these 100's and 100's of processes don't always get cleaned up properly, especially if the software doesn't clean up after it's fork() children.
My question is: Is there a tidy command-line thing I can do to kill off these processes without having to either reboot, or do it 1-by-painstaking-1? All processes are assoc.'d with a particular uid. So, I'm guessing there's a nice sequence of piped commands of the form:
ps -all processes | grep to a uid | kill
or maybe there's a script out there?
Thank you!
Bryan C.
Not surprisingly, these 100's and 100's of processes don't always get cleaned up properly, especially if the software doesn't clean up after it's fork() children.
My question is: Is there a tidy command-line thing I can do to kill off these processes without having to either reboot, or do it 1-by-painstaking-1? All processes are assoc.'d with a particular uid. So, I'm guessing there's a nice sequence of piped commands of the form:
ps -all processes | grep to a uid | kill
or maybe there's a script out there?
Thank you!
Bryan C.