WAS THIS PAGE HELPFUL? Leave Feedback
With the increased availability of large systems with typically 8 or more processors, 16 GB of memory and terabytes of disk space, it becomes necessary to consider re-configuring GroundWork Monitor Enterprise to take advantage of these resources in large deployments. GroundWork Monitor Enterprise was developed and optimized for dual core 4-8 GB systems and the usual optimizations (threadpools, heap space allocations). As such, it won't take full advantage of a large system with the default deployment.
This document describes the steps necessary to reconfigure GroundWork Monitor Enterprise for large systems. This configuration can be used with systems that have:
- 16 GB or more of memory
- 8 or more processor cores (for bare metal installations - VMWare instances require further consideration - 2 cores will offer better performance)
- The GroundWork disk space requirements for this type of installation do not change significantly
|This configuration requires version 7.1.1 or 7.1.0 and newer. Please do not apply any of these changes to previous releases. It will probably break and won't be supported.|
We separate the UI and the backend into separate JVMs, each of which can be given several gigabytes of RAM. Some service ports change so that the two JVMs have their own places to do business.
- GroundWork Monitor 7.1.1 or 7.1.0 installation successfully operating.
- Message preprocessing setup according to the Bookshelf page Preprocessing Log Messages. We suggest the pre-processing of LogMessages by Operation Status for the default views. This reduces load significantly on the PostgreSQL database instance.
- The hardware requirements for this capability are the same or better as described in the overview.
- You need nagios user privilege.
Please reserve at least an hour downtime for performing this installation and have all the users logged out of the GroundWork portal.
|Before you install|
To install on a GroundWork server where the single-instance JBoss AS system is deployed, do the following:
- SHMMAX SHMALL Kernel parameters - Possibly edit /etc/sysctl.conf to allow shared memory to be assigned up to the amount available in the system. Here we show typical entries which work for a system with 16 GB of RAM, allowing up to 1/2 the RAM to be allocated as shared. Use this as a starting point, remembering that allowing the OS to assign 100% of physical RAM as shared memory will likely cause some performance issues.
Once this file is edited use the following command to make the changes immediately effective (there are run time methods but we want the changes to be permanent as well as immediate):
- , then ssh to postgresql server and edit:
and change the setting as follows (use 1/4th of your physical memory to determine your shared_buffers):
and restart postgresql:
- Run the installer script as the root user (or user with root sudo rights), and the installer will guide you with the available options:
DO NOT run this script as the nagios user.
|Configuration Information (after successfully converting to dual JBoss):|
The following command will now show 3 JVMs running: 2 JBoss instances, and 1 Tomcat instance.
If you want to revert back to single-instance JBoss AS configuration, go to the folder where you have saved the 7.1.1.single-jboss.backup.tgz or 7.1.0.single-jboss.backup.tgz the file and run the following commands sequentially:
|Adjust the following command if using version 7.1.0 replacing 7.1.1.|