We have SAS EBI on a windows server. The server admin is worried about the amount of SAS process, some of them were started weeks ago. We decided to try to schedule a restart of the Object Spawner, every Sunday morning, so all the SAS process spawned would be kill. It worked for some of them. Here is our batch file:
net stop "SAS [SASCONFIG-Lev1] Object Spawner"
TIMEOUT /T 60 /NOBREAK
net start "SAS [SASCONFIG-Lev1] Object Spawner"
Are missing something ?
Is there another way to clean up those zombie process ?
Sometimes a client is disconnected from a Workspace Server process in a way that causes it to hang completely. Normal termination signals will not work at all. On UNIX, this means to use kill -9, which causes the kernel to deallocate the process without any cooperation from the process itself.
There must be a similar radical command in Windows.
When the processes are real zombies they will not go away when the initiator (object spawner) is gone.
The root cause can be a hangup in the closing part.
Similar radical commands exist in Windows PsTools (TechNet) PSkill -Taskkill lseems to be a normal interenal command. On my W7 it is giving a reaction. /F as -9 switch. If you are worried think on restarting all serverice and validate the machine is clean and setup on a regular planned schedule moment.
There are a lot of sasprocesess that are services (multithreaded).