Summary
After a period of successful operations and running multiple policies, the ECX appliance becomes unresponsive. Restarting the ECX appliance resolves the issue.
Symptoms
By default, Linux limits the number of file descriptors per process to 1,024. After reaching the limit, the new process/thread is blocked and the ECX appliance may become unresponsive.
In this situation, the Virgo log /opt/virgo/serviceability/logs/log.log on the ECX appliance generates the following error:
Encountered an error executing the step java.lang.OutOfMemoryError: unable to create new native thread
Resolution
The ECX appliance relies on the underlying Linux operating system settings as well as preset ECX appliance properties. In some cases, you may need to optimize default Linux system settings or ECX properties in order to improve performance.
To ensure optimal ECX appliance performance, you can optimize a number of concurrent processes within the ECX appliance by changing the content of the configuration file and restarting Virgo:
- Edit the file /opt/virgo/repository/ecx-usr/com.syncsort.dp.xsb.service.jobmanager.properties by logging in to the appliance as root. Contact Catalogic Software Data Protection Technical Support to obtain your root password and for information about accessing the ECX virtual appliance command prompt.
- Edit the values of the parameters to the following:
- jobmanager.taskExecutorPoolSize=500
- jobmanager.maintenanceTaskExecutorPoolSize=2
- jobmanager.threadsPerNode=100
- jobmanager.threadsPerJob=100
- jobmanager.skipLimit=100
- Restart Virgo on the appliance through the following commands:
- service virgo stop
- service virgo start