RMA Architecture Question

[lollbackj]

Currently deciding the correct architecture for implementing Advanced Host Monitor in a national organisation with major offices in 5 Australian cities (Sydney, Melbourne, Adelaide, Brisbane and Perth).

All offices are connected by a private internal network with link speeds of between 10 and 40Mb.

The preferred solution is to have a single instance of AHM and to deploy two agents at each site (primary and backup) to perform tests. Installation of an agent on every server isn't an option at this time.

WAN outages are rare but not unheard of so I'm looking for advice on the following items:

• Is Active or Passive the correct solution?
  Is a load balanced configuration appropriate (between the two agents at each site)

All thoughts/assistance greatly appreciated.[/list]

[KS-Soft]

Is Active or Passive the correct solution?

If connection between networks brakes often for short time, its better to use Active RMA. Also it may help when some antivirus/firewall cannot handle frequent incoming connections properly.
Otherwise any agent is good (e.g. if network problems occur rarely but require 5 or more minutes to be fixed).

Is a load balanced configuration appropriate (between the two agents at each site)

Sure, why not.

Regards
Alex

[lollbackj]

So what would be the 'best practice' configuration in this instance:

• Every server being monitored has a ping test at 15s intervals using the primary Active RMA for 'test by'.
  All other rules for a server (disk space, count files, event log etc)use the ping test as a master test.

Should a new test be created to test for Active RMA and this test become a master test for each of the ping tests to ensure that ping tests and all subsequent tests are suspended if connectivity is lost?

[KS-Soft Europe]

It's good idea to use Ping test with rma itself string (instead of target host name like localhost) as Master test to all tests, performed by particular RMA agent.
Quote from manual:
http://www.ks-soft.net/hostmon.eng/rma- ... htm#rmacfg

- Active RMA: Things a little more tricky when you need to check Active RMA status. Yes, you may setup the same "Ping localhost" test to check agent status. However this may lead to some delay in reaction as HostMonitor will not perform test if agent was connected but lost connection just a moment ago. HostMonitor may wait up to several minutes for new connection before changing test status to "Unknown". Dependant test items will be delayed as well so such delay is not a big problem, you will not receive a lot of alerts. However if for some reason you need to receive alert immediately, there is solution: setup Ping test using Active RMA and type rma itself string instead of localhost or target host name. In such case HostMonitor will display agent status immediately. HostMonitor will not use backup agent (if any) when specified agent is not connected; also HostMonitor will not wait for agent re-connection.

[KS-Soft]

It's good idea to use Ping test with rma itself string (instead of target host name like localhost) as Master test to all tests, performed by particular RMA agent.

Not in this case.
When you have 2 agents that check the same set of servers (using backup or load balancing mode), you do not want to stop monitoring when 1 agent is out of service. Otherwise what sense of using backup agent??
You may setup such test to be informed when there are some problems with agent. However for Master test you should use some "regular" test that can be backuped by backup agent just like Dependant tests.

In other words, if you have 2 RMA agents that checks many hosts in the same LAN, I would recommend to setup several Master tests (each test should be performed by RMA, not HostMonitor)
1) main Master test (TestA) that will check that at least 1 RMA works: "Ping localhost"
2) subMaster tests dependant on TestA, should check if target systems are alive. Setup 1 Ping test for each monitored system (TestB1, TestB2...)
Then setup all other items, make them dependant on TestB1, TestB2...

Regards
Alex