Avoid file server for more IO operations

Summary: By default, IO operations are serialised through the file server, which can add significant overhead. We can bypass this using the `raw` option or the `prim_file` module, depending on context. `wa_raft` already uses this in many places, but here we enable it in a few more to further boost performance.

Reviewed By: hsun324

Differential Revision: D78900618

fbshipit-source-id: 597baa97a66bb803f4f41f307d6e9ca2a011c507

Author: TD5

src/wa_raft_storage.erl

@@ -730,7 +730,7 @@ handle_delayed_reads(#state{queues = Queues, module = Module, handle = Handle, p
 -spec handle_create_snapshot(SnapshotName :: string(), Storage :: #state{}) -> {ok, wa_raft_log:log_pos()} | {error, Reason :: term()}.
 handle_create_snapshot(SnapshotName, #state{name = Name, path = Path, module = Module, handle = Handle, position = Position} = State) ->
     SnapshotPath = filename:join(Path, SnapshotName),
-    case filelib:is_dir(SnapshotPath) of
+    case filelib:is_dir(SnapshotPath, prim_file) of
         true ->
             ?RAFT_LOG_NOTICE("Storage[~0p] skips recreating existing snapshot ~0p.", [Name, SnapshotName]),
             {ok, Position};
@@ -746,7 +746,7 @@ handle_create_snapshot(SnapshotName, #state{name = Name, path = Path, module = M
 -spec handle_create_witness_snapshot(SnapshotName :: string(), Storage :: #state{}) -> {ok, wa_raft_log:log_pos()} | {error, Reason :: term()}.
 handle_create_witness_snapshot(SnapshotName, #state{name = Name, path = Path, module = Module, handle = Handle, position = Position} = State) ->
     SnapshotPath = filename:join(Path, SnapshotName),
-    case filelib:is_dir(SnapshotPath) of
+    case filelib:is_dir(SnapshotPath, prim_file) of
         true ->
             ?RAFT_LOG_NOTICE("Storage[~0p] skips recreating existing witness snapshot ~0p.", [Name, SnapshotName]),
             {ok, Position};
@@ -800,7 +800,7 @@ cleanup_snapshots(#state{path = Path}) ->
 %% Private functions
 -spec list_snapshots(Path :: string()) -> [{wa_raft_log:log_pos(), file:filename()}].
 list_snapshots(Path) ->
-    SnapshotNames = filelib:wildcard(?SNAPSHOT_PREFIX ++ ".*", Path),
+    SnapshotNames = filelib:wildcard(?SNAPSHOT_PREFIX ++ ".*", Path, prim_file),
     Snapshots = lists:filtermap(fun decode_snapshot_name/1, SnapshotNames),
     lists:keysort(1, Snapshots).