We have hangfire deployed in production and are seeing odd behavior, probably due to our not fully understanding the implications of our production infrastructure yet. Here is the scenario.
We have 2 zones, each with multiple servers where Hangfire runs. We have a single DB shared between zones. We have 1 zone that is hot and one cold. What we are seeing is that there are jobs being picked up by machines in the cold zone, even after we have disabled the jobs via config and restarted the application pool.
I’m mostly interested in any best practices for this sort of set up. I’m thinking either queues or zone specific job ids may work but does anyone have any practical advice for keeping jobs from running on old code in a blue-green deployment scenario?
Thanks,
Brett