Clearing Zombie processes

Folks,

Our setup has a lot of processes going zombie every now and then (mostly Call Object Kernels). Also what we observe is when the zombie processes increase the app becomes slow as well. Is there any relation?

Also, can we manually delete the zombie processe through the Web SAW (for Enterprise Server there is an option to Delete/Remove the zombie processes). Is that safe?

Normally we have to bounce JDE Services to cleanup all processes.

Please advise

Thanks,
CNC Guy
E1 8.11 SP1
UNIX

The zombie processes hold on to some memory so that could be causing some of the slowness depending on how many you have on there. I've clearing zombies through Web Saw for years and never had an issue with it. It works great.

Thanks cncjunior. I think we will use this to clear our zomibies as well. But we still have to figure out the reason for those zombies. I see a lot of these errors in kernel logs which go to zombie :

"UEUE04950060-Fetch failed for _01_6013 in F986130"

Any clues?

Thanks,
CNC Guy

Its looking for a queue and failing.

From the initial underscore it looks as if there may be a duff entry in the F986130.

Try (during downtime), making a note of all the queues, regenning the F986130 and adding them all again.

As I've stated a few times on these forums, nearly all zombies are caused by database issues.

Can't agree with you more clark. But the problem is how to prove that it is because of the DB. As i said in our case the DBA's are from some other company and hence the tussle of proving where lies the issue

This forum could use a good discussion on zombies, heck even Hitler was pissed about it.

I see zombied kernels as the result of the code choking on anything, not just database issues. Sometimes they have an equivalent "Web Exception" error on the client side, other times not. Bad code, etc.

I am going to start a thread dedicated to zombies. Just make sure you have your zombie outbreak survival kithandy.


[ QUOTE ]
Its looking for a queue and failing.

From the initial underscore it looks as if there may be a duff entry in the F986130.

Try (during downtime), making a note of all the queues, regenning the F986130 and adding them all again.

As I've stated a few times on these forums, nearly all zombies are caused by database issues.

[/ QUOTE ]

Usually, we delete the Zombie process by Server Manager. There is one option for Enterprise Server - option to Delete/Remove the zombie processes.
We not encounter any issue after deleting it.

Thanks.

you need to look for the cause of zombie process. In my case there are few BSFN that also caused zombie on server when executed WebMethods interface.
you can note down the zombie process id and look for the same in "net kernel" logs ,this will give you info on the Application that cause zombie process.

Hello List,

A document I have with me .