We are working with a DMA failover pair: dma1a and dma1b.
- A Windows service currently runs on
dma1aand pushes data into the DataMiner system via the Data API. - For resiliency, we are considering deploying a reverse proxy to both servers, using a reverse proxy setup to route traffic to whichever DMA is active.
- The DMA pair does not use hostname failover but is configured with a Virtual IP (VIP) that fails over between the two nodes.
Our questions:
- We would have a separate service which pushes to the VIP. The VIP would route to the reverse proxy which would be installed on every DMA and the reverse proxy sends the data to localhost:34567. Is this approach achievable?
- If this approach is not supported or reliable, what is the recommended way to programmatically determine which DMA in the pair is currently active, so the service can dynamically direct data to the correct agent?
Chris Glover [DevOps Advocate] Answered question 11th April 2025
If you're using VIP based fail over, you don't need a reverse proxy, just make your VIP the target of the sending application. It'll always goes to the active DMA
If you use hostname based failover, you'll need a reverse proxy in front.
We've used ha-proxy for this, as it can determine the active host (inactive host stops listening on the port)
Chris Glover [DevOps Advocate] Edited answer 11th April 2025