High availability service

The ingress service provides a highly available endpoint for the Ceph Object Gateway. The ingress service can be deployed to any number of hosts as needed. IBM recommends having at least two supported Red Hat Enterprise Linux servers, each server configured with the ingress service. You can run a high availability (HA) service with a minimum set of configuration options. The Ceph orchestrator deploys the ingress service, which manages the haproxy and keepalived daemons, by providing load balancing with a floating virtual IP address. The active haproxy distributes all Ceph Object Gateway requests to all the available Ceph Object Gateway daemons.

A virtual IP address is automatically configured on one of the ingress hosts at a time, known as the primary host. The Ceph orchestrator selects the first network interface based on existing IP addresses that are configured as part of the same subnet. In cases where the virtual IP address does not belong to the same subnet, you can define a list of sub-nets for the Ceph orchestrator to match with existing IP addresses. If the keepalived daemon and the active haproxy are not responding on the primary host, then the virtual IP address moves to a backup host. This backup host becomes the new primary host.