|Access||Nagios can Ping Client||Nagios can reach TCP/6556 on Client||Nagios can SSH to Client||Client can SSH to Nagios||Can Install Software|
Host that you can only ping
Set the Agent type to Ping-Only
Make sure you have no "manual checks" that apply to ALL_HOSTS
Check_MK will follow the rules you create so it would try to fetch the data for your manual checks. Avoid this by making them apply only to the systems that have the "tcp" aux tag.
Host that you can't ping
Create a new Agent type called noping
Create a check_dummy active check
Assign this as the host check for hosts with the "noping" Agent type.
For non-WATO users:
Host where you can't reach port 6556
Can you reach SSH instead?
Create a new Agent type called sshpull
Use a datasource_program that calls ssh for connecting to the client
Let it use SSH Force-Command to run the agent
If you can still ping, don't modify the host check. Otherwise, create a different host check.
Host where you can't reach ssh but that could reach the nagios server
Use the mechanism of datasource_programs, i.e. if the clients can reach the nagios server via SSH, you could use check_mk_remote to push results back.
Create a new Agent type called sshpush
Create a datasource that reads in the Agent result file for sshpush
Create a different host check that interprets the result file.
Host where you can't install software but reach ssh
Modify the agent script to be run via SSH on the host - so i.e. instead of running
ssh remote-host check_mk_agent
ssh remote-host "content_of_check_mk_agent"