Notification jobs, scheduled task jobs and application jobs run in the TaskProxy.exe executable, which does the following:
Every minute, all InTrust servers update the LastServerPing column of the ADCServer table of the configuration database with the current date/time.
An UPDATE trigger on the LastServerPing column executes a Stored Procedure which checks the date/time in the LastServerPing column against the current date/time:
This stored procedure also closes any sessions that are open (running) on the failed InTrust server - this is to prevent jobs managed by other InTrust servers waiting for a job managed by the failed server to complete.
Every 60 seconds, a specially intended InTrust Server Failure Detection data source executes a script which checks the status of each InTrust server and generates an event for each InTrust server that is down (with status of 0).
The InTrust Server is Down real-time monitoring rule detects these event(s) and invokes the response action (once per event).
After an InTrust server failure has been detected, the failover process switches from the failed server to the standby server, as follows:
For details about failover configuration, see Backup Procedures for InTrust.