mongo server is stop, Reconnect Increase delay

[panzixin39]

mc_worker.erl
init()->
......
Error ->
receive after 3000 -> ok end, %%add
proc_lib:init_ack(Error)
end.

[comtihon]

why should we wait after error was detected?

[panzixin39]

Control the frequency of reconnection，Reducing CPU Occupancy

[comtihon]

I think some top supervisor should be responsible for worker restart delay. Not the worker itself, as it know nothing about will it be restarted or not.

[panzixin39]

Reconnection requires delay。Where to change, you are free

[comtihon]

it is not supported out of the box.
There are several options:

• You should use your own supervisor, have mc_worker be a child and restart it with backoff (can be increment). You can try something ready like this or this
• use standard supervisor, but custom start function - wrapper around mc_worker with try-catch and your delay code. I think it is the easiest and the fastest solution.
• use standard supervisor + standard worker, but without reconnect. Use your monitor process, which will monitor the pool and shoot a timed restart (start new worker actually) if error encountered.

[comtihon]

I think reconnect with delay is useful, but I can't promise I'll have a time implementing this in the nearest future.

[panzixin39]

ok.