Load balancer setup
This section describes the recommended approach for Core to monitor, load balance, and set up automated detection of a Primary Core failure and failover to the paired Secondary Core.
Figure 1. Load balancer setup for Core monitors
At a high level the load balancer should perform the following:
-
Probe the status of the Primary server by issuing /status/status.html and receiving the “MOBILEIRON-STATUS: OK” message. This indicates a healthy Core. Any other response should be considered a failure.
-
Probe the status and state of the Secondary by issuing /hastatus.html and receiving the “MOBILEIRON-STATUS: OK (Mode: Standby)” response. The “OK” indicates the server is up and running. The “Mode: Standby” indicates the Core is in Standby mode and it is safe to failover traffic in case of a Primary Core failure. The other possible value for “Mode” is “Primary” which indicates the server is acting as Primary.
-
The load balancer should be set up to keep traffic routed to the Primary server and only failover traffic when the following conditions apply:
-
Primary probe fails and
-
Standby Core shows “Mode: Standby”
-
-
The failover should NOT resume traffic back to the Primary when the Primary becomes available (Manual Resume = Yes). Resuming traffic back to the Primary is a manual process.
Important: If the option to swap Core roles is implemented as part of the Core HA Recovery, the necessary changes must be made to the load balancer to match the Core HA architecture, that is, having the right monitors checking the correct Cores.