Currently each worker heartbeat is hardcoded to have a floor of 250ms and ceiling of 1 minute. This doesn't work well for applications as they might want to enforce quicker detection of crashed workers than every minute.
What we should do is extend remus::worker::Worker to be more like remus::server::Server so that people can determine better heart-beat controls for the worker. Something like the following should suffice:
This would construct a worker and state it should heartbeat no faster than every 50msec, and no slower than every 150msec. If the worker doesn't heartbeat every 150msec the server will presume it to be dead.
Currently each worker heartbeat is hardcoded to have a floor of 250ms and ceiling of 1 minute. This doesn't work well for applications as they might want to enforce quicker detection of crashed workers than every minute.
What we should do is extend
remus::worker::Worker
to be more likeremus::server::Server
so that people can determine better heart-beat controls for the worker. Something like the following should suffice:This would construct a worker and state it should heartbeat no faster than every 50msec, and no slower than every 150msec. If the worker doesn't heartbeat every 150msec the server will presume it to be dead.