OEM alert on Zombie process

 

We have got below OEM alerts for RAC1.COM Sever

Message=Internal error detected: java.lang.Throwable:oracle.sysman.gcagent.tmmain.execution.LongOpManager$ZombieDetection:1138.

13c agent oracle.sysman.gcagent.task.TaskFutureImpl$WrappedTask:621

Please follow the below steps :

 

Note: Backup AGENT_INST/sysman/config/emd.properties

Run the following command Using EMCTL where you are facing issue related to Zombie process:

Note:  Stop the agent and perform the activity

./emctl stop agent

 

1. Set the following zombies property at agent side

$/AGENT_INST/bin/emctl setproperty agent -allow_new -name _zombieSuspensions -value true

$/AGENT_INST/bin/emctl setproperty agent -allow_new -name _canceledThreadWait -value 900

$/AGENT_INST/bin/emctl setproperty agent -allow_new -name _zombieThreadPercentThreshold -value 0

$/AGENT_INST/bin/emctl setproperty agent -allow_new -name _zombieCreateIncident -value false

$/AGENT_INST/bin/emctl setproperty agent -allow_new -name _cancelThreadTimeout -value 300

$/AGENT_INST/bin/emctl setproperty agent -allow_new -name _traceSeverityForTypes -value noTypes

 

Please find the below output

[oracle@rac bin]$ emctl setproperty agent -allow_new -name _zombieSuspensions -value true

Oracle Enterprise Manager Cloud Control 13c Release 3

Copyright (c) 1996, 2018 Oracle Corporation. All rights reserved.

EMD setproperty succeeded

 

OR

Go to agent location and edit emd.properties and f

Edit AGENT_INST/sysman/config/emd.properties file and add following parameters

_traceSeverityForTypes=noTypes

_zombieSuspensions=true

_canceledThreadWait=900

_zombieThreadPercentThreshold=0

_zombieCreateIncident=false

_cancelThreadTimeout=300

 

Explanation of each terms:

O _canceledThreadWait == > The amount of time before a canceled thread is considered a 'zombie' thread

O _zombieSuspensions ==> Whether to indefinitely suspend a zombie task

O _zombieThreadPercentThreshold ==> The number of 'zombie' threads (as a percent of the REGULAR thread pool) before a bounce will be started; set to zero to disable auto restarts

Note: Setting these parameters , agent restart due to zombies would reduce/stop and incident dump shall get created

O _zombieCreateIncident - Whether to create an incident for the zombie abort

 

2. Backup file <AGENT_INST>/sysman/config/s_jvm_options.opt and add following parameter at EOF

-XX:+UseMembar

 

3. Also, I would like to suggest you to increase the memory values set at agent side:

from: (default)

agentJavaDefines=-Xmx128M -XX:MaxPermSize=128M

 

to:

agentJavaDefines=-Xmx512M -XX:MaxPermSize=128M

 

4. Restart agent

./emctl start agent

 

5. Monitor and check status