differences in installation styles between nimldr and the rpm installer

Products

• DX Unified Infrastructure Management (Nimsoft / UIM)

Issue/Introduction

We are  deploying some new hubs on a new Linux deployment and some questions have arisen about why there are two installers (nimldr and rpm) and which one is the right one to use, and why they seem to behave differently.

The rpm complains about needing chkconfig and initscripts but then it doesn't appear to actually use them, it uses the systemctl / systemd method of installing the nimbus service.  So why does it want those packages?  

Which one is correct, /etc/init.d or systemd?

Which installer should I use?

Environment

DX UIM - Any version

nimldr installer

.rpm package for robot

 

Resolution

It mostly comes down to "style" in terms of how things are coded but essentially - both nimldr and the rpm should "try" to use the newer systemd/systemctl style for the services, but if this isn't available, /etc/init.d/ is used.

The rpm requires the chkconfig and initscripts packages to be present up front, "just in case" the /etc/init.d stuff is needed whereas the nimldr doesn't check for this (and then may subsequently fail to install correctly if it's not present).

Additionally, the way they handle /etc/init.d is a little different - nimldr doesn't create a link in /etc/init.d unless needed, whereas rpm creates kind of a fallback/dummy link.

The robot installed with the rpm will continue to function if /etc/init.d/nimbus is removed as long as systemd/systemctl are still available.

Otherwise there should be no fundamental differences between the behavior of the robot/services with either installer type.

Officially, nimldr is considered the only "correct" way to install a standalone Linux hub, as the rpm can only install a robot, but either way works to install a robot and both are considered fully supported.