TIBCO has introduced a new feature starting with LMI 6.2.0 to better assist EVA users with properly setting resource configurations for various LogLogic LMI components. There are actually profiles for hardware appliances too but users won't interact with them as much as they will the VM-related resource profiles so this article does not focus on the hardware profiles. The profile system is not and cannot address every scenario. Users may still need to make some tweaks to the settings' values. However, the default values provide a great start. For example, a large number of advanced searches or advanced searches over large time frames returning a large amount of data may necessitate increasing the Data node's memory allocation above the default value even though the settings for other components would not need to change. That is a change that could be required with or without the resource profiles being available as a feature however, the procedure for implementing the change is what differs if you are using LMI 6.2.0 or higher because resource profiles would be in effect.

First and foremost is that the resource profiles are automatic and no user intervention is necessary to enable them, implement them or manage them unless you need to customize the settings managed by them. Creating your own profile is not supported; the design is intended to only support editing the existing profiles. Since most environments are covered by the default values there will be no need in those circumstances to touch the profiles' configurations. Not every LMI setting is controlled by the profiles and those settings that are controlled by profiles are only indirectly controlled by virtue of being specified in the configuration files that are defined in the profiles.

The profile settings are considered dynamic at the granularity of system reboots due to a Linux startup script being used to choose the proper profile and enforce it at boot time. The startup script performs the same checks every boot to determine which profile should be used. Again, this occurs without the need to explicitly enable them. If the VM's resource allocations are modified at the vSphere level then upon the next reboot the EVA's memory/CPU profile may be different if the change in resources was sufficient to trigger a different profile. Otherwise the various component settings managed by the profiles will not ever change from one bootup to another without either manual intervention or a change to the VM's resources.

All EVA profiles are stored in /loglogic/conf/vm_profiles. There are currently 5 profiles available in 6.2.0 and more may be created in the future:
16gb_4cores
256gb_Xcores
32gb_Xcores
64gb_Xcores
128gb_Xcores

The profiles can be based on memory and CPU cores with memory as the first criteria. As of LMI 6.2.0 the profile selection based on CPU core quantity is not yet available hence all non-default profile names use X in place of the core quantity. The default profile is 16gb_4cores.

There are 5 files in each profile:
capability.xml
my.cnf.LXVirtual
my.cnf.MXF
my.cnf.MXVirtual
my.cnf.STVirtual

Any setting configured in any other file will not be controlled by the profiles. At boot time the /etc/rc3.d/S94pre_logappd init script will determine which profile to apply then apply it by copying the files from the profile directory to /loglogic/conf. It will pick the best profile that fits based on memory and CPU (future) cores with memory always being the primary criteria.

The settings that vary across the profiles are currently memory allocations for the following components:
Aggregation node
Correlation node
Data node
Query node
Zookeeper
Indexer
Tomcat
MySQL InnoDB buffer pool
MySQL Key buffer size

Any edits to the active copies of the profile-related files in /loglogic/conf/ will be overwritten by the original file from the profile in /loglogic/conf/vm_profiles/ when the appliance reboots because that's when the profile's files are copied over to /loglogic/conf/ to become the active files.

Therefore, if you have already made your change to the standard configuration file in /loglogic/conf because that's traditionally how the changes were applied when profiles didn't exist then, in addition to that, you'll need to make the same changes to the copy of the file in the dynamic profile directory corresponding to the active profile in order to make them survive a reboot. Be sure to edit the file for the correct profile based on memory and CPU cores. Engine/mtask restarts are only necessary when editing the active copy in /loglogic/conf.

As a reminder, if you sufficiently change the memory or CPU core allocations for the EVA to cause another profile to be used then your custom changes will not be used. So you will need to apply your previous custom changes to the new profile that is in effect.

Starting in LMI 6.2.0 there is a new dynamic resource profiles feature that helps to dynamically allocate various resource settings for multiple application components. This article provides a brief overview of how they work. However, this article is not a how-to for editing the profiles. For general guidance on how to edit resource profile settings refer to article 36251.

The memory configuration chart for the various profiles and application components is documented in the LogLogic LMI EVA 6.2.0 Quick Start Guide located here. Refer to article 36251 for a general description of how to make profile setting changes.