Fix issue where merge attempts to sync deleted hint

wday-contrib 1702

Author: aef-

src/bitcask.app.src

@@ -1,6 +1,6 @@
 {application, bitcask, [
     {description, "Yet another key/value storage engine"},
-    {vsn, "2.1.0.3"},
+    {vsn, "2.1.0.4"},
     {modules, []},
     {registered, []},
     {applications, [kernel, stdlib]},

src/bitcask.erl

@@ -1,7 +1,7 @@
 %% -------------------------------------------------------------------
 %%
 %% Copyright (c) 2010-2017 Basho Technologies, Inc.
-%% Copyright (c) 2018-2023 Workday, Inc.
+%% Copyright (c) 2018-2024 Workday, Inc.
 %%
 %% This file is provided to you under the Apache License,
 %% Version 2.0 (the "License"); you may not use this file
@@ -3019,14 +3019,14 @@ no_pending_delete_bottleneck_test2() ->
         [begin
              ?assertEqual(ok, bitcask:delete(B1, K))
          end || {K, _} <- Data],
-        bitcask:merge(Dir1),
+        ok = bitcask:merge(Dir1),
 
         bitcask:iterator(B2, -1, -1),
         put_till_frozen(B2),
         [begin
              ?assertEqual(ok, bitcask:delete(B2, K))
          end || {K, _} <- Data],
-        bitcask:merge(Dir2),
+        ok = bitcask:merge(Dir2),
         bitcask:iterator_release(B2),
 
         %% Timing is everything. This test will timeout in 60 seconds.
@@ -3282,7 +3282,7 @@ leak_t1() ->
     [bitcask:delete(Ref, <<X:32>>) || X <- lists:seq(1, DelKeys)],
     io:format("After deleting ~p keys, lsof says: ~s", [DelKeys, Used()]),
 
-    bitcask:merge(Dir),
+    ok = bitcask:merge(Dir),
     io:format("After merging, lsof says: ~s", [Used()]),
 
     io:format("Q: Are all keys fetchable? ..."),
@@ -3618,7 +3618,7 @@ no_crash_on_key_transform_test() ->
                      default_dataset()),
     bitcask:list_keys(B),
     bitcask:fold(B, fun(K, _V, Acc0) -> [K|Acc0] end, [], -1, -1, true),
-    bitcask:merge(Dir, [{key_transform, CrashTx}]),
+    ok = bitcask:merge(Dir, [{key_transform, CrashTx}]),
     ok = bitcask:close(B),
     B2 = bitcask:open(Dir, [{key_transform, CrashTx}]),
     ok = bitcask:close(B2),

src/bitcask_fileops.erl

@@ -1,7 +1,7 @@
 %% -------------------------------------------------------------------
 %%
 %% Copyright (c) 2010-2017 Basho Technologies, Inc.
-%% Copyright (c) 2022-2023 Workday, Inc.
+%% Copyright (c) 2022-2024 Workday, Inc.
 %%
 %% This file is provided to you under the Apache License,
 %% Version 2.0 (the "License"); you may not use this file
@@ -407,6 +407,8 @@ read(#filestate { fd = FD }, Offset, Size) ->
 
 %% @doc Call the OS's fsync(2) system call on the cask and hint files.
 -spec sync(#filestate{}) -> ok.
+sync(#filestate { mode = read_write, fd = Fd, hintfd = undefined }) ->
+    ok = bitcask_io:file_sync(Fd);
 sync(#filestate { mode = read_write, fd = Fd, hintfd = HintFd }) ->
     ok = bitcask_io:file_sync(Fd),
     ok = bitcask_io:file_sync(HintFd).

src/bitcask_merge_delete.erl

@@ -1,6 +1,7 @@
 %% -------------------------------------------------------------------
 %%
 %% Copyright (c) 2012-2017 Basho Technologies, Inc.
+%% Copyright (c) 2024 Workday, Inc.
 %%
 %% This file is provided to you under the Apache License,
 %% Version 2.0 (the "License"); you may not use this file
@@ -182,7 +183,8 @@ delete_files(Files) ->
 -ifdef(TEST).
 
 multiple_merges_during_fold_test_() ->
-    {timeout, 60, fun multiple_merges_during_fold_body/0}.
+    [{timeout, 60, fun multiple_merges_during_fold_body/0},
+    {timeout, 60, fun multiple_merges_during_fold_body_with_corrupt_hintfiles/0}].
 
 multiple_merges_during_fold_body() ->
     Dir = filename:join(?TEST_FILEPATH, "bc.multiple-merges-fold"),
@@ -212,7 +214,7 @@ multiple_merges_during_fold_body() ->
     PutSome(),
     Count1 = merge_until(Dir, 0, CountSetuids),
     PutSome(),
-    bitcask:merge(Dir),
+    ok = bitcask:merge(Dir),
     PutSome(),
     merge_until(Dir, Count1, CountSetuids),
 
@@ -223,8 +225,44 @@ multiple_merges_during_fold_body() ->
 
     ok.
 
+multiple_merges_during_fold_body_with_corrupt_hintfiles() ->
+    Dir = filename:join(?TEST_FILEPATH, "bc.multiple-merges-fold-hintfile"),
+    B = bitcask:open(Dir, [read_write, {max_file_size, 50}]),
+    PutSome = fun() ->
+                      [bitcask:put(B, <<X:32>>, <<"yo this is a value">>) ||
+                          X <- lists:seq(1,5)]
+              end,
+    PutSome(),
+    PutSome(),
+    Bstuff = get(B),
+    FoldFun = fun(_K, _V, 0) ->
+                      receive go_ahead -> ok end,
+                      1;
+                 (_K, _V, 1) ->
+                      1
+              end,
+    _SlowPid = spawn(fun() ->
+                            put(B, Bstuff),
+                            bitcask:fold(B, FoldFun, 0)
+                    end),
+    CountSetuids = fun() ->
+                           Fs = filelib:wildcard(Dir ++ "/*"),
+                           length([F || F <- Fs,
+                                        bitcask:has_pending_delete_bit(F)])
+                   end,
+    PutSome(),
+    _DeletedHints = [file:delete(H) || H <- filelib:wildcard(Dir ++ "/*.hint")],
+    _Count1 = merge_until(Dir, 0, CountSetuids),
+    PutSome(),
+    ok = bitcask:merge(Dir),
+
+    ok = ?MODULE:testonly__delete_trigger(),
+    0 = CountSetuids(),
+
+    ok.
+
 merge_until(Dir, MinCount, CountSetuids) ->
-    bitcask:merge(Dir),
+    ok = bitcask:merge(Dir),
     Count = CountSetuids(),
     if (Count > MinCount) ->
             Count;
@@ -571,7 +609,7 @@ fork_merge(H, Dir) ->
 merge_these(_H, TestDir, Ids) ->
     Files = [lists:flatten(io_lib:format("~s/~w.bitcask.data", [TestDir,Id])) ||
                 Id <- Ids],
-    bitcask:merge(TestDir, [], Files).
+    ok = bitcask:merge(TestDir, [], Files).
 
 incr_clock() ->
     bitcask_time:test__incr_fudge(1).