cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 

Sending oneagent traffic through F5 load balacer

alejandro_herna
Organizer

We have an enviroment
where we are monitoring 30 hosts, we have F5 load balancer and we would like to
load balance the Oneagent traffic to Dynatrace Managed Server without
installing environment ActiveGate, how can I set the agent to send traffic to
the virtual IP or domain of the F5 load balancer?


6 REPLIES 6

There is no such option to force oneagent send data though some other ip than the one belongs to ActiveGate or Cluster.

Sebastian


Julius_Loman
Leader

You are monitoring just 30 hosts. Do you have a multi-node Dynatrace cluster? If not, what is the purpose of putting an F5 for load balancing? You effectively cannot manage the endpoint addresses the agent is connecting to. This can be only by done mangling your /etc/hosts or DNS) as the list of endpoint URLs is sent to OneAgents by Dynatrace server.

However, Dynatrace handles load balancing pretty solid by itself, so I don't think you need a load balancer at all between agents and cluster.

Unless you are forced to do the content inspection (which would be much simpler to configure with using HTTP proxy anyway) I'd definitely not recommend to put any components into the communication.


Thank you @Julius L. and how is the communication handled for cluster ActiveGates when enabling Mobile RUM? described in scenario 3 on documentation where load balancers are in place:

https://www.dynatrace.com/support/help/deploy-dynatrace/managed/installation/managed-deployment-scenarios/


You can have LB before ActiveGate if you have more than one. Then you will put proper ip address of LB inside mobile RUM configuration. But Generally LB is not mandatory there, you can have single AG and make it public on port 9999.

Sebastian


Actually, I don't think you are able to specify multiple endpoint URLs for the beacons for Mobile RUM in case you have multiple cluster activegates. Thus you will need to somehow load-balance the connections or do at least the fail-over on server side. (Beacon URL gets inserted into the application at instrumentation time).


This is what I've had in mind 🙂

Sebastian