This PR relates to vernemq/vernemq#988 issue, and assumes
that erlio/gen_server2#1 pull request was merged.
Internal gen_server2 message queue can grow infinitly
when broadcast rate reaches a certain level and eventually
consume all available system memory. In order to mitigate
this problem process can drop i_have messages under
heavy load. Messages of that kind are retried if
not acknowlaged, so it doesn't break algorithm correctness.
By default messages aren't dropped, this feature has to be
configured.
This PR relates to vernemq/vernemq#988 issue, and assumes that erlio/gen_server2#1 pull request was merged.
Internal gen_server2 message queue can grow infinitly when broadcast rate reaches a certain level and eventually consume all available system memory. In order to mitigate this problem process can drop i_have messages under heavy load. Messages of that kind are retried if not acknowlaged, so it doesn't break algorithm correctness. By default messages aren't dropped, this feature has to be configured.