From 85ac519c899d98338a1818c5bcceafe71ba11b15 Mon Sep 17 00:00:00 2001
From: Karl Nilsson <kjnilsson@gmail.com>
Date: Fri, 8 Nov 2024 10:28:42 +0000
Subject: [PATCH] Ignore unexpecte files in the log directory.

some filesystems (e.g. NFS) create hidden files to deal with deletion
of open files. The presence of a file: .nfs000000000000000 would stop
the osiris log from initialising.

This commit updates the code to ignore such files and proceed anyway.
---
 src/osiris_log.erl        | 15 +++++++++------
 test/osiris_log_SUITE.erl | 22 ++++++++++++++++++++--
 2 files changed, 29 insertions(+), 8 deletions(-)

diff --git a/src/osiris_log.erl b/src/osiris_log.erl
index 2ca2b79..fc2a816 100644
--- a/src/osiris_log.erl
+++ b/src/osiris_log.erl
@@ -1829,7 +1829,10 @@ orphaned_segments([<<Name:20/binary, ".index">>,
     %% when we find a matching pair we can return
     Acc;
 orphaned_segments([<<_:20/binary, ".segment">> = Dangler | Rem], Acc) ->
-    orphaned_segments(Rem, [Dangler | Acc]).
+    orphaned_segments(Rem, [Dangler | Acc]);
+orphaned_segments([_Unexpected | Rem], Acc) ->
+    %% just ignore unexpected files
+    orphaned_segments(Rem, Acc).
 
 first_and_last_seginfos(#{index_files := IdxFiles}) ->
     first_and_last_seginfos0(IdxFiles);
@@ -2038,15 +2041,15 @@ overview(Dir) ->
             {Range, EpochOffsets}
     end.
 
-index_files_with_segment([<<_:20/binary, ".segment">> | Rem], Dir, Acc) ->
-    %% orphaned segment file, ignore
-    index_files_with_segment(Rem, Dir, Acc);
+index_files_with_segment([], _, Acc) ->
+    lists:reverse(Acc);
 index_files_with_segment([<<Name:20/binary, ".index">> = I,
                           <<Name:20/binary, ".segment">>
                            | Rem], Dir, Acc) ->
     index_files_with_segment(Rem, Dir, [filename:join(Dir, I) | Acc]);
-index_files_with_segment(_, _, Acc) ->
-    lists:reverse(Acc).
+index_files_with_segment([_OrphanedOrUnexpected | Rem], Dir, Acc) ->
+    %% orphaned segment file or unexpected file, ignore
+    index_files_with_segment(Rem, Dir, Acc).
 
 
 
diff --git a/test/osiris_log_SUITE.erl b/test/osiris_log_SUITE.erl
index 33d1821..91e7eb8 100644
--- a/test/osiris_log_SUITE.erl
+++ b/test/osiris_log_SUITE.erl
@@ -89,8 +89,9 @@ all_tests() ->
      small_chunk_overview,
      overview,
      init_partial_writes,
+     init_with_unexpected_file,
      overview_with_missing_segment,
-     overview_with_missing_segment_at_start
+     overview_with_missing_index_at_start
     ].
 
 groups() ->
@@ -1791,6 +1792,23 @@ run_scenario(Config, NumChunks, MsgSize, Scenario) ->
 
     ok.
 
+init_with_unexpected_file(Config) ->
+    Data = crypto:strong_rand_bytes(1500),
+    EpochChunks =
+        [begin {1, [Data || _ <- lists:seq(1, 50)]} end
+         || _ <- lists:seq(1, 20)],
+    Dir = ?config(dir, Config),
+    Log = seed_log(Dir, EpochChunks, Config),
+    osiris_log:close(Log),
+    Segments = filelib:wildcard(filename:join(Dir, "*.segment")),
+    Indexes = filelib:wildcard(filename:join(Dir, "*.index")),
+    ?assertEqual(2, length(Segments)),
+    ?assertEqual(2, length(Indexes)),
+    ok = file:write_file(filename:join(Dir, ".nfs000000000000000000"), <<"bananas">>),
+    _ = osiris_log:init(?config(osiris_conf, Config)),
+    ?assertEqual({{0,999},[{1,950}]}, osiris_log:overview(Dir)),
+    ok.
+
 overview_with_missing_segment(Config) ->
     Data = crypto:strong_rand_bytes(1500),
     EpochChunks =
@@ -1821,7 +1839,7 @@ overview_with_missing_segment(Config) ->
                         filename:join(?config(dir, Config), "*.index")))),
     ok.
 
-overview_with_missing_segment_at_start(Config) ->
+overview_with_missing_index_at_start(Config) ->
     Data = crypto:strong_rand_bytes(1500),
     EpochChunks =
         [begin {1, [Data || _ <- lists:seq(1, 50)]} end