locked
Best practice for handling HAProxy Failover RRS feed

  • Question

  • What is currently the best practice way of handling failover for HAproxy behind the Azure Load balancer?

    We currently have a two-node setup of HAProxy using keepalived on the nodes to detect and failover virtual IP addresses. Are there any solutions in i.e. Azure Load Balancer that can handle floating IP and handle the failover outside of the actual HAProxy nodes?
    Monday, November 25, 2019 12:26 PM

Answers

  • The template I linked above was designed by someone on the Networking team. 

    You can also check out this GitHub Issue of someone else who implemented HAProxy without an Azure Load Balancer, instead going for a programmatic switch of IPs on failover. 

    Tuesday, November 26, 2019 9:41 PM

All replies

  • Here is an example template that uses an Azure Load Balancer and HAproxy with 2 VMs

    This also uses a floating ip. Can you clarify what exactly you need with the failover? 

    Monday, November 25, 2019 6:56 PM
  • Is it possible for you to provide any information in regards to the recommended solution from Microsoft in regards to HAProxy behind Azure Load Balancer?

    Is it possible to have a floating IP that i.e. Azure Load Balancer manages and ‘routes’ to a HAProxy?

    We currently have a solution where the HAProxy nodes handles failover internally using keepalived, and we are looking for a solution where we place that logic within the Azure platform. So if there is a Azure Service that can handle the failover, any information that you can provide about that would be most helpful.

    Tuesday, November 26, 2019 8:40 AM
  • The template I linked above was designed by someone on the Networking team. 

    You can also check out this GitHub Issue of someone else who implemented HAProxy without an Azure Load Balancer, instead going for a programmatic switch of IPs on failover. 

    Tuesday, November 26, 2019 9:41 PM