HPUX version B.11.23 as reported by uname -a

Start with default signal I believe is -15
#kill PID
Process still present.
#kill -9 PID
Process still present, even after repeated attempt and waiting.

PPID becomes 1 which is the init process
started at boot time.


So far only way to stop has been to reboot host.

Any suggestions, HP support has been unhelpful.

Unkillable processes are usually trying to complete a system call. The process will die once the system call completes. If you use glance you can see what the process is waiting for. If the process is inappropriately unkillable, the culprit is usually a defective driver.

Thanks for response.
The processes that I want to kill have been hanging one one host since I posted the question. Waiting has not been the issue. I should be able to kill any process, regardless of what it may be waiting for. I have not ever had this kind of problem with same code ported to SGI IRIX or RedHat Linux.

I do not know what glance is , but will try to find it.
Most of the unkillables are ssh connections that scripts are using to collect
information from redundant hosts in a group. But I have had instances where
scripts and C compiled binaries have overloaded a system causing me to
attempt cleanup by killing them, but they will not die... thereby mu only recourse was to reboot. These same programs are then fine after reboot.

I was curious as to whether there was a known hpux kernel bug that I could repair.

No you shouldn't. As one example, mkdir() requires several io operations. If it starts but does not complete, the file system is in an illegal state.