Hello, is it possible to un-connect only one agent from the collector and not like the restart of the collector un-connecting all of the agents? The problem was that one agent connected with its AGENTNAME_, and I could not stop/start the agent itself as this is running with customers data at the moment (it's a webserver). So I restarted the collector and then the issue with the wrong name AGENTNAME_ was gone, but all other agents also reconnected. So for this my question, is it possible to only un-connect and re-connect one agent triggered this operation from the collector without having to stop/start the agent?
You should not be worried about XXX_1 instant because it will be disappeared after 72 hours by default and collector restart should not be recommended in this case. We can restart the web and host agent from the agent views dashboard.
If you want to remove the agent permanently then web/app restart is mandatory (also if you want to release the agent) and to stop collecting data we can uncheck the capture event from the system profile > agent group > agent mapping for that specific agent.
Some background - that 1 is the indexing of the agent which occurs when two agents of the same name connect at the same time. Often it is actually the case of an application restarting and not fully disconnecting before the agent is back up again (so the collector thinks two are connected at once).
To avoid this there is actually a debug flag the you can set on the collector which is pretty common:
To address your actual question - really in order for any change to the agent name to occur the agent will need to reconnect to the collector, either via a collector restart or an application restart. Since the agent is running "within" the server process there is no way to restart it besides restarting the application (with the exception being host agents and the primary web server agents which can be restarted on their own). There is no way I am aware of to reconnect one particular agent from a collector.
Many thanks Babar and James for your replies 🙂
72h waiting is too long as if you have an own dashlet created for a specific agent name, this is no more updated and shows nothing. This is why I restarted the collector so that the concerned (and unfortunately all others) agent has to reconnect and its service was not interrupted for our customers.
Hi James, or any other,
I tried the debug flag on our Test Center Edition and no second instance  was created. If I create this rule now onto our Production Edition, does this has no other side effects?
There shouldn't be any real noticeable or significant effects - a lot of deployments make this the standard and we have it in production with no issues.
The only thing that can occasionally be a minor issue that I have seen is some of the monitoring dashboards such as the topology view which lists all of the processes don't know how to represent agents that seem to have identical names which is possible if there are two sub web server agents under the same primary and do not have an ApacheNodeName configured. It causes it to aggregate these into one line with the suffix (x2) which doesn't really make sense if you drill down and have only agent's data displayed.
This is a very specific case and as long as you're aware of it is not an issue but I wanted to mention it.