I have a scenario where the agent should be deployed on a remote machine (that cannot be accessed from the orchestration server - but there is access from the agent to the orchestration server).
Did someone get this sort of agent up and running (the server is located in AWS, the agent is located in my office network).
What do you mean by "not accessible from the automic server"?
I believe that you can make the agent work if your AE server can communicate with the agent.
Which basically mean you have to open some port in your firewall, 2217 and 2300 to be precise.
You can check this discussion for more detail:
Connections in an AE environment
Was the information provided by Dat helpful in shedding light to your questions? I am not sure how/what your network communication policy is like, but for the agent, it would need communication both way for it to work properly. One thing to consider is looking at the proxy component which may help your scenario.
You need to open at least the control port (8871) from your AE to your agent.
The last time I meddled with AWS, one could configure one's own firewall rules and open up these ports.
(edit: I just realized your server is in AWS and your agent is in your office ... well, unless you have a VPN between you and your AWS instance (not sure if that's even possible), that wouldn't work unless you expose the agent's control port to the internet so your AWS instance can reach it via a public IP. In that case, you'd also probably be dealing with NAT, and the AE doesn't do NAT, so it needs to be told about the true IP address of the agent via the .ini file. While this can be done, I for one would not like the security implications of exposing an Automic agent to the internet.
In short, it works if you establish a TCP connection, but it may not be a good idea.
Hi Carsten, thanks for the information!
I'll update when i get it working!