sap host agent spawning zombies

[Former Member]

I have installed sap hostagent 720 patch 194 on HP-UX ia64 / Oracle 11g systems. It's having the same problem on all of our servers. When I start the host agent, within seconds it starts spawning zombies. I know it to be the host agent as I perform this...

ps -ef |grep def

    root 29961 28836  1 10:01:00 ?         0:00 <defunct>

    root   304 28836  1 10:02:01 ?         0:00 <defunct>

    root 29186 28836  0 09:59:58 ?         0:00 <defunct>

Then look for the parent:

ps -ef |grep 28836

    root 29961 28836  1 10:01:00 ?         0:00 <defunct>

    root   304 28836  1 10:02:01 ?         0:00 <defunct>

    root   558 28836  0 10:03:02 ?         0:00 <defunct>

    root 29186 28836  0 09:59:58 ?         0:00 <defunct>

    root 28836     1  0 09:58:57 ?         0:00 /usr/sap/hostctrl/exe/saposcol -l -w60 pf=/usr/sap/hostctrl/exe/host_profile

I get this same thing on all of my systems. I have been trying to find a solution for this for some time so I hope someone here can help.

Thanks in advance,

Kyle

[Former Member · Accepted solution]

It appears that the problem is related to our version of the VM-ware we are using, we are now in the process of upgrading our systems to include VM-ware.  Thanks for all the replies.

[rafaelgalindo · Accepted solution]

Kyle,

Please, which version you have installed for solution this problem?

[former_member182657]

Could you confirm current SAP Host agent version after upgrade ?

[Former Member]

I will certainly try upgrading again. I've already upgraded the hostagent at least once, maybe twice since this problem started.

[former_member182657]

Hi Kyle,

When I start the host agent, within seconds it starts spawning zombies. I know it to be the host agent as I perform this...

I've already upgraded the hostagent at least once, maybe twice since this problem started.

I don't think here an issue with sap host agent,as you've already upgraded it twice & got same results,I suspect your system became victim of zombies here & assuming your system is completely open to the outside world ( No use of firewall or any type of port security ). Kindly confirm if 'm wrong.

As you said spawning Zombies after few secs,then here you need to check for unwanted entries under cron jobs of the system by checking crontab -e.

Here i could be wrong but i suggested all because now we need to consider every aspects to resolve the issue.Addition to it,are you getting this type of behavior on single system ? If yes then try to perform the same activity on other to confirm.

Hope you would consider my reply.

Good luck !!

[Former Member]

Thank you for your suggestions, but our systems are secured from the outside world by extensive firewall systems so I'm confident that this is not the issue. And this is happening on all of our HP-UX systems in the Sandbox, Dev, QA, and Prd environments.

It was suggested that the problem had to do with the VMware we were using which would make sense except that two of the systems don't use VMware.

[former_member182657]

Thanks for the information.

root 29961 28836  1 10:01:00 ?         0:00 <defunct>
    root   304 28836  1 10:02:01 ?         0:00 <defunct>
    root 29186 28836  0 09:59:58 ?         0:00 <defunct>

Have you tried killing the PID 28836 with kill -9 <pid> & rebooted the server ? If not try to kill the process & restart the complete system to check again.

Also suggest you to refer links Zombie process - Wikipedia, the free encyclopedia

for <defunct> status.

Hope this will help you.

[Former Member]

Yes, I can kill the processes individually, but if I just use saphostctrl -stop, it'll kill the offending parent and all child processes go away even though they are defunct.

[former_member182657]

Hi,

but if I just use saphostctrl -stop, it'll kill the offending parent and all child processes go away even though they are defunct

It could be,but as far as i know these are the processes which get detached from their parent & then sitting into the process table at OS level,as per OS standards system tries to clean these type of processes in a defined time manner,but fails to clean out completely (An OS expert or OS support can guide you better then mine explanations on this).

So as per my suggestions here we could do two things,first try to restart the complete system & check again

Second is to update the current kernel with the latest patch available with again complete restart of system.

You could refer SAP Note  1246652 - Child processes of a work process remain as zombie processes

Hope this will help you.

Good luck !!

[manoj_nayak2]

Hello Kyle,

Can you please try to download and upgrade Host control to latest version (SAP HOST AGENT 7.20 SP205  / 7.20 SP206 ) and let us know result.

./saphostexec -upgrade

Regards,

[Former Member]

Hello Kyle,

Please check with your Operating system team if this issue exists for other os processes which are started in the server.

Also please try to install n-1 version of hostagent and see if the same issue is happenning.

Do let me know your findings.

Thanks,

Sowmya