Using Load Balancer VIP with ActiveGates for SNMP polling and traps

[TahirJanwari]

Hello Community,

We monitor around 600+ network devices in Dynatrace using both SNMP polling and SNMP traps through Three ActiveGate servers.

Our challenge is that when an ActiveGate is replaced or decommissioned, we may need to update SNMP settings, trap destinations, firewall rules, or allowed SNMP manager IPs on many network devices.

We are considering using a Load Balancer VIP in front of the ActiveGates.

The idea is:

• Devices send SNMP traps to one VIP.
• Dynatrace ActiveGates are in the backend pool.
• For SNMP polling, we want the devices to see one stable polling source IP, either through VIP/NAT or a similar design.

Has anyone implemented this type of setup?

I would like to know:

1. Does this work for Dynatrace SNMP traps?
2. Can it also work for SNMP polling, so devices see one stable source IP?
3. Should source IP be preserved for traps?
4. Any issues with SNMPv2c, SNMPv3, duplicate traps, polling failures, or failover?
5. What is the best practice to avoid updating hundreds of devices when ActiveGate IPs change?

Any guidance or real-world experience would be appreciated.

Thank you.

[Mohamed_Hamdy · Accepted solution]

Hi Tahir,

A VIP can potentially be used for SNMP traps, as discussed in the following link:

https://community.dynatrace.com/t5/Extensions/VIP-for-SNMP-Trap/m-p/243298

However, SNMP polling operates in the opposite direction, from the ActiveGate to the network device. Therefore, a standard inbound load-balancer VIP would not address the polling requirement.

If the objective is for all polling ActiveGates to appear to the network devices as a single permitted SNMP manager IP, a stable outbound SNAT address could be considered. Alternatively, the network devices could allow a dedicated and stable ActiveGate subnet.

NAM may also work through a stateful SNAT configuration, since the checks are initiated from the Synthetic-enabled ActiveGate toward the monitored targets. However, this should be validated in the target environment.

In summary, a single load-balancer solution would not cover both traffic directions:

• An outbound SNAT could be considered for SNMP polling and NAM.
• A separate Layer 4 UDP VIP could be considered for SNMP traps.

These approaches are technically feasible in principle, subject to the correct network configuration. For the trap VIP in particular, the original device source IP should be preserved, and each trap should be forwarded to only one ActiveGate to avoid incorrect source identification or duplicate processing.

As neither approach is clearly documented by Dynatrace as an officially supported reference design, I recommend validating both through a controlled POC and confirming the proposed design with Dynatrace Support before production deployment.

Kind regards,
Hamdy