Hello all and happy new year.
I have an HP-UX system running HP-UX B.11.31 U ia64 1161301486 and it has a weird process called "svhptdaemon" that's using 100% of the CPU. Anyon knows what this process is? I've tried googling the answer but even Google doesn't know 
CPU TTY PID USERNAME PRI NI SIZE RES STATE TIME %WCPU %CPU COMMAND
1 ? 28 root 152 20 72K 64K run 23992:09 100.40 100.23 svhptdaemon
Any help would be very much appreciated.
Can you find the executable? The path might give you a clue. Sadly I do not have an HP-UX server available any more.
I found these two hits looking for just "HP UX SVHPT". Do they help?
Robin
Thanks but it doesn't really help. Such a weird process. I cannot stand HP-UX. 
I can't kill this process either. Kill nor Kill -9 options work. If I run which svhptdaemon - no entries. Rebooting the server, the process starts up again. Must find it!
svhpt is part of the vmunix kernel on HP. Since it has been a loong time since I did HP, we really should ask somebody like Methyl or Made in Germany.
That explains why it cannot be killed.
While I have never met a svhptdaemon myself on HP-UX 11.31. Maybe it is associated with a certain function.
The kernel processes have PID 0. Can be listed with
pgrep -l -P 0
or
ps -ef | awk '$3==0'
Is there high %sys in sar 1 100 or vmstat 1 ?
Are there kernel messages in /var/adm/syslog/syslog.log ?
Thanks "jim mcnamara" and "MadeInGermany" for the responses.
The service PID is 28 and the PPID is 0.
I had checked syslog but nothing related to that specific process was in the log file.
both SAR and vmstat shows about 50% usage for sys:
10:11:27 %usr %sys %wio %idle
10:11:28 0 50 0 50
10:11:29 0 50 0 50
10:11:30 0 50 0 50
10:11:31 0 51 2 47
procs memory page faults cpu
r b w avm free re at pi po fr de sr in sy cs us sy id
2 1 0 127867 5713748 0 0 4 0 0 0 0 394 849 51 0 50 50
2 1 0 127867 5713663 0 0 1 0 0 0 0 457 850 49 0 50 50
2 1 0 127867 5713663 0 0 0 0 0 0 0 449 697 49 0 50 50
How do you attribute the 50% sys usage to that one process? That is for all processes is it not? BTW - when ANY process runs it spends time in kernel code (sys) when invoking a syscall, .e.g. open(). That means vmstat and sar would normally show a summed % sys for all processes system wide.
Doesn't mean that the daemon is or is not using most of that %.
My mistake. I was trying to say the same thing you just did. The usage is around 50%.
Using tools such as lsof or truss , if you have them installed.
truss -rall -wall -p PID
Examine the calls, perhaps some file is missing and program runs continuously in loops or similar issues (could explain the large %sys usage)
Also issue a find to perhaps to locate from which script is that spawned as rbatte suggested.
Something in the line of :
find /sbin/init.d -type f -exec egrep "svhptdaemon|svhpt" {} /dev/null \;
If you get a match, examine that script for possibility to disable that and what that is actually.
Perhaps i can offer more assistance in couple of days when i get to hpux box.
Hope that helps
Regards
Peasant.
Unfortunately I don't have lsof nor truss installed on this box.
I issued the find command but it came back with no results.
---------- Post updated at 04:04 PM ---------- Previous update was at 03:57 PM ----------
I decided to run find on the entire / fs. This came up, not sure what to make of it. This came from a file in the /tmp directory. /tmp/toolset/spin.0509
Running Threads (TSRUNPROC) and idle Processors
===============================================
TICKS TICKS I TICKS
SINCE SINCE C SINCE NREADY
TID PID PPID RUN IDLE PRI SPU STATE S MIGR FR LO AL COMMAND
------- ----- ----- --------- --------- ---- --- ----- - --------- -- -- -- -------
9822392 9922 9804 0 414 231 0 SYS ? 1104611 7 0 -- crashinfo
Processor #0
TICKS TICKS I TICKS
SINCE SINCE C SINCE NREADY
TID PID PPID RUN IDLE PRI SPU STATE S MIGR FR LO AL COMMAND
------- ----- ----- --------- --------- ---- --- ----- - --------- -- -- -- -------
57 28 0 425067099 425067099 152 1 SYS ? 425067116 0 7 -- svhptdaemon
Interesting! Run
fuser /tmp/toolset/spin.0509
Does it state a PID? Of what process?
It only returned the directory name
root@server:/
> fuser /tmp/toolset/spin.0509_1552
/tmp/toolset/spin.0509_1552:
root@server:/
>
I can only guess that shptdaemon belongs to some clustering software.
And it is certainly misbehaving.
Please contact HPE support! Or Oracle if you run the Oracle RAC software.
Thanks all for your input. I'll reach out to HPE support and post back with the fix (provided it can be).