23 Aug 2019 05:51 AM - last edited on 11 Apr 2023 05:50 AM by Karolina_Linda
We have an environment where Agents can connect directly to the Dynatrace-cluster (by interest proxy) or thru their environment ActiveGate.
In the environment to put it simple we have an agent on host A and B, an application on A sends web requests to B. So in normal operation there is a pure path between app A and app B.
1) if a request failes to connect to B (example: socket time-out), this will be reported as a error on service " unmonitored host"
2) what happens if the order changes e.g. Agent B sends his info before A, or there is a large delay between A and B (temp network problem on B)
3) when the network is unavailable between for Dynatrace components, every measure is recovered but not the network measures from the monitored hosts ( there was no issue on the network there)
Is this explainable?
Solved! Go to Solution.
Actually it doesn't have any different if agents are connected directly or via proxy to DT because network is measured on monitored hosts itself.
1) yes, sometimes dynatrace is creating as well opaque service which is representation of unavailable entity that agent is trying to reach (but it's not unmonitored because DT knows this IP / hostname)
2) Should work the same way. It depends of transactions itself. If network is open / performance well only in one direction, part of transactions may work normally, part not. For affected processes you should see connectivity drops.
3) You are talking about network issues between Dynatrace and OneAgent? After reconnection cached metrics were sent but not those related to network, yes? If network issues were related to host that is monitored by OneAgent, maybe there were some issues with measurements. Hard to say. You can always download support archive from server and check logs.
Great answer, thx!