Question

I have several web-servers and need them to use Quartz. The clustering feature of Quartz would be ideal, but it requires that the servers clocks are in complete sync. They have a very scary warning about this:

Never run clustering on separate machines, unless their clocks are synchronized using some form of time-sync service (daemon) that runs very regularly (the clocks must be within a second of each other).

I cannot guarantee complete clock synchronization, so instead of using the clustering feature I was thinking to have a single Quartz instance (with a stand-by for fail-over). Having a single instance executing jobs is not a problem, but I still need all of the web servers to be able to schedule jobs.

Can I directly add jobs into the JDBCJobStore from the web servers, and will they be picked up by the (non-clustered) Quartz server? I would be doing this by creating schedule instances in the web servers to add jobs. These instances would never be started, just used to access the JobStore.

Answer 1

Wrote a test program that creates a “non-clustered” Quartz scheduler using the same JobStore as the “real” scheduler (also non-clustered), and schedules jobs. After a few seconds, these jobs do get executed, so it seems to work.

Update: I cross-posted this question to the Quartz forum, and got the answer that this should work. In a related question they state that

The jobs can be inserted into that database by another process by:

1- using the rmi features of quartz from another process, and using the quartz API

2- instantiating a scheduler within another process (e.g. webapp), also pointing it to the same database, but NOT start()ing that scheduler instace, and the using the quartz api to schedule jobs.