My 4 servers are running scheduler, but 5th started crashing with "There's already an active RQ scheduler" message. But there is no active RQ scheduler on this particular machine. There is active scheduler on other machine.
I've inspected the code and I do not understand what for are registering birhts and deaths. There is lock already implemented. The scheduler_key is shared between all instances. There is no real difference between "local" and "remote" instances.
I think that briths and deaths can be removed. Any thoughts?
Edit: sometimes rqscheduler runs on two or more machines, sometimes just one can work at once. This is bad, because there is no redundancy (failover)
Hi.
My 4 servers are running scheduler, but 5th started crashing with "There's already an active RQ scheduler" message. But there is no active RQ scheduler on this particular machine. There is active scheduler on other machine.
I've inspected the code and I do not understand what for are registering birhts and deaths. There is lock already implemented. The scheduler_key is shared between all instances. There is no real difference between "local" and "remote" instances.
I think that briths and deaths can be removed. Any thoughts?
Edit: sometimes rqscheduler runs on two or more machines, sometimes just one can work at once. This is bad, because there is no redundancy (failover)