Archived
This forum has been archived. Please start a new discussion on GitHub.
Server go out of control when icegridnode shutdown abnormally
Hi, I am a new guy here , hope someone can help me with my problem.
I developed a system using icegrid, but when I ran some test, I found something unexpected:
when I shutdown the icegridnode process with shell command "kill -9", or under certain condition the icegridnode process went down abnormally, the servers running on this node, their parent process ID became 1! Then I started the icegridnode again, I can't find any running server with the IceGridGUI tool, but they were still running indeed, just that they went out of control of icegridnode!
Is that expected behavior? What can I do in this situation?
I developed a system using icegrid, but when I ran some test, I found something unexpected:
when I shutdown the icegridnode process with shell command "kill -9", or under certain condition the icegridnode process went down abnormally, the servers running on this node, their parent process ID became 1! Then I started the icegridnode again, I can't find any running server with the IceGridGUI tool, but they were still running indeed, just that they went out of control of icegridnode!
Is that expected behavior? What can I do in this situation?
0
Comments
-
Hi,
Yes, this is expected behavior. If you kill the parent process, the the child processes will be re-assigned to the process group of the init process (with pid 1). You need to kill the icegridnode process group if you want the processes forked by the icegridnode to be killed as well. See here for some information on how to kill the process group.
I don't recommend killing the IceGrid node, the IceGrid node should always exit in a timely manner if the servers it forks also exit in a timely manner. If a server mis-behaves and hangs on exit, you can kill it using the IceGrid administrative clients.
Cheers,
Benoit.0 -
OK, thank u very much!0
