Ask HN: Java Scheduler in a cloud environment
Quartz doesn't promise a lot, but it also appears low-level enough for one to build on top of it chaining, failover, and building one's own execution context. These are not pros, but rather a statement that its drawbacks can be worked around. Not looking forward to having to include JDBC job store (already have another DB-backend for other stuff) and a time-sync service just to get clustering up but I'm not surprised that it is required either.
Obsidian on the other hand seems more feature complete, I'm not sure why no-code, no-xml job configuration is spouted so strongly by its advocates as a feature as I'd rather be able to version control things but one appears to be able to do everything programmatically if one wishes. The wiki isn't complete, http://obsidianscheduler.com/wiki/Admin_Job_History#Job_History_Details and unlike Quartz there seems to be fewer resources available online on "How to do X in Obsidian".
The licensing fees aren't too enticing either.
The dream scheduler would support failover out-of-the-box, and be designed for high availability. It would have a REST API (Obsidian has it, one can be built on top of Quartz), it should be possible to chain jobs, and it should also be able to consume system events so that one could phrase jobs such as "run this job sometime after midnight as long as this event has happened"