Issue: Due to build promotion servers get restarted and there were on job running on one server. Since it has recover flag true ,when servers came up ,this job started running on all server.
Note usually one job runs on only one server but above case is only happens when we restart all servers.
We have three clusters and each has one JVM .Quartz scheduler is on each JVM.
org.quartz.jobStore.clusterCheckinInterval = 20000
1.6 is very very old. Anyway to upgrade to at least 1.6.6 (which is also very old, but the most stable in the 1.6.x line).
Are you sure with your build promotion you didn't have a problem with the quartz configuration such that a different config (such as the one built into the quartz jar to use RAMJobStore) was picked up?
Are you sure that it is the recovery trigger that fired the job? (and if so, how)?