We have a need to add custom information about the network devices between two monitored servers. The first step required to start it is understand from the agent plugin which machines are being accessed by it.
We may also use SolarWinds to retrieve the network device information but we need to understand the topology itself. I thought we could maybe run a tracert script for each accessed node and utilize its information to add to the custom devices.
So to summarize:
- first obstacle is - how can we know which hosts are accessed by the oneagent running a oneagent plugin.
- what do you think about the overall approach
In this case, you will have to write ActiveGate plugin to create such entities. Dynatrace detects the communication passes such device (on layer 4 or above) you will see it as a load balancer icon in the service flow with the IP address. If you create a custom device with the same IP addresses, it will automatically correlate. For the custom device created you can send additional metadata and metrics.
Tracert, however, might reveal other network nodes that are doing just layer 3 communication such as routers / switches (IP only). You won't see them in the service flow or in the purepaths and adding such nodes won't add them to the service flow or correlate data from them.
From oneagent plugins you cannot create custom devices nor see the communication, unless you will do it on your own by running netstat or ss.
Hi @Julius L.,
The target devices are indeed network devices such as routers and switches that will be revealed by tracert for example. My thought was to use tracert to reveal and update the custom devices with the network dependencies, hence, maybe adding some more insights into DAVIS for root cause analysis.
We do have solarwinds as well but I'm not sure yet if it has any mapping of dependencies for communication purposes.
You certainly can add those custom devices, but I don't think the AI will pick up them when analyzing problems from deep monitoring data from OneAgents. This is valid for devices that are not seen on the OSI layer 4 or above. Normally there is no information in the IP packet payload which network devices did the packet pass, so the correlation is limited when the main source is the data from service requests (PurePaths).
For devices that can be seen (for example in X-Forwarded-For headers), the correlation should happen.
In general you will not see those devices as taking part in transactions (there is no agent there, they are not entry points for transactions or their IP rather are not called inside transactions). This is how network devices works, they are transparent from transaction point of view. You will see custom network devices in dynatrace, you can even collect some metrics using ActiveGate from them (if they have such option via API) and push them to dynatrace. But this will be something not correlated in transactions.