I am trying to find some documentation on some detail on fault tolerance with Camunda.
Specifically timers … If Camunda server goes down before a scheduled timer event goes off - will it be triggered when the Camunda/tomcat server is brought back up ?
The short answer is yes. A timer is interpreted as guaranteed not to fire before the due date. A timer is implemented as a job. hence it is stored in the database as a job with a future due date. Thus as long as you have not lost the database, the job persists.
Jobs are picked up and run by the job executor. Hence as soon as an engine node with a job executor is available, the timer job will likely be run…
Perhaps this reference [1] may give you confidence.