[ddorian43]

But cron isn't highly available/distributed with consistent state/scheduling

[stephenr]

Run cron on all your boxes, have them pull a job at a time from the database. Or shit, go full tilt with a clustered Redis job queue.

All of these are better options than "embed the entire job info in a URL and hand it off to a remote scheduling service".

[scarface74]

http://mikehadlow.blogspot.com/2012/04/database-as-queue-ant...

[stephenr]

The user doesn't need a queue though. They just need a process to find posts with a publish date of <= now that aren't published and publish them. That's a one line indempotent sql update statement in any decently architected application.

[scarface74]

This was the other part of the comment I replied to:

Run cron on all your boxes, have them pull a job at a time from the database. Or shit, go full tilt with a clustered Redis job queue.

You're solving for the example but the original poster noted multiple jobs.