Discussion found here has been locked and this is why I have opened this one up.
Consider the following scenario:
You are coming from a single Orion instance which had Orion agents in use. Most of the agents deployed were set with agent initiated communication (passive). The remaining agents which were deployed in DMZ networks were set with server initiated communication (active).
After some period of time, the SolarWinds Platform grew to include a High availability server. This created a challenge, since monitoring servers by means of agents from an HA pool with two unique IP addresses meant that you had to manually reconfigure the 'passive' installed agents in order for them to communicate back to the VIP of the HA pool. This is the only way that 'passive' agents can successfully communicate back to the SolarWinds Platform in case of a failover event. Changing what IP the 'passive' agents communicate back to is not easy task. You have the option to do it manually on you 250 servers, or you can do it through group policy using the Orion Agent Active Directory Group Policy Administrative Template. You come learn that the GPO Template method works, but not always, since an agent service restart is sometimes necessary. Furthermore, this method only works for Windows computers that are joined to the domain. The 90 odd servers out of the 250 servers total which are Linux based, must be reconfigured manually.
You consider briefly changing the polling method of these deployed agents to server initiated, thinking this would solve all your problems by giving you access to all agents, regardless of which polling engine in the HA pool is the active one. You have a look around in the Manage Agents view as well as online documentation and quickly find, that changing the communication method from passive to active is not possible. You try redeploying the agent from the Add Agent page, disregarding the warning about the agent being reinstalled and you end up having duplicate agents in the Solarwinds Platform. You find out that there is no option to group agents by agent mode in the Deploy Agent on Network view, so you can't really tell what mode existing DMZ and non-DMZ Agents node are using.
You then reconsider this method and choose to manage, administer and update the agents to communicate with the HA Pool VIP manually. A week later after several false alerts and missing KPI's due to maintenance mode being set on Nodes, all you 'passive' are now communicating with the HA Pool VIP.
A year passes and your network grows in number of servers. The Solarwinds Platform grows as well, this time with two Additional Polling Engines and two HA pools. The plan is now to monitor the additional servers. You have 700 servers total that you need to monitor. 250 of these servers are being monitored by the existing HA Pool either actively or passively. How to monitor the remaining 450 servers with 4 polling engines (2x active HA + 2x passive HA) will take some time to consider. You immediately think back to the time you considered changing the agent mode on all monitored servers to 'active' and decide that this is the way to go forward. You then poke holes in your firewalls, change polling method from agent to WMI/ICMP and loose important resources and their historical KPI's from SAM, SCM etc. while doing so, set all servers to unmanaged, uninstall all 'passive' agents from the Manage Agents view, then push a new agent install using your third party software deployment tool along with a settings file that configures the agents to run in active mode (msiexec /i SolarWinds-Agent.msi /qn TRANSFORMS=SolarWinds-Agent.mst).
Lastly, you now need to reconnect to these 'active' agents in a fast and hassle free fashion, change polling method back to agent based and hopefully not run into any issues that would require you to delete and re-add nodes.
What, in your opinions, should be the next step and what could have been done differently (bar agent communication mode from the get go) to avoid all these hurdles?
