| Overview |  |
riak_ensemble_peer to handle
asynchronous work such as K/V gets and puts.
Long-lived worker process used by riak_ensemble_peer to handle
asynchronous work such as K/V gets and puts. The peer maintains a pool
of workers and routes requests to workers based on object key, ensuring
that operations that affect the same key are serialized on the same worker.
Thus, workers also enable partitioned queuing. These workers can also be
paused and resumed. This is used by the peer when necessary to enforce
necessary barriers. For example, a leader will pause all workers before
attempting to change ensemble membership, then resume workers after. This
prevents workers from issuing requests during a changing ownership set,
since those requests will likely be rejected.
| loop/1 | |
| maybe_pause/1 | |
| pause_workers/2 | |
| start/1 | |
| unpause_workers/2 |
loop(ETS) -> any()
maybe_pause(ETS) -> any()
pause_workers(Workers::[pid()], ETS::ets:tid()) -> ok
start(ETS::ets:tid()) -> {ok, pid()}
unpause_workers(Workers::[pid()], ETS::ets:tid()) -> ok
Generated by EDoc