Bug: Error on CubDB startup

[TomHoenderdos]

Suddenly I saw CubDB had crashed this morning, this was the error returned to me. After I cleaned the cubdb files in the data_dir this error was gone. I wasn't able to download the invalid file, so I'm not giving you much to work with I'm afraid.
If this happens again I will try to fetch the invalid db file and post it here.

  {:error,
  {%ArgumentError{
    message: "errors were found at the given arguments:\n\n  * 1st argument: invalid external representation of a term\n"
  },
  [
    {CubDB.Store.CubDB.Store.File, :raise_if_error, 1,
     [
       file: 'lib/cubdb/store/file.ex',
       line: 156,
       error_info: %{module: Exception}
     ]},
    {CubDB.Btree, :new, 2, [file: 'lib/cubdb/btree.ex', line: 64]},
    {CubDB, :init, 1, [file: 'lib/cubdb.ex', line: 1232]},
    {:gen_server, :init_it, 2, [file: 'gen_server.erl', line: 851]},
    {:gen_server, :init_it, 6, [file: 'gen_server.erl', line: 814]},
    {:proc_lib, :init_p_do_apply, 3, [file: 'proc_lib.erl', line: 240]}
  ]}}
  ** (EXIT from #PID<0.4989.0>) shell process exited with reason: an exception was raised:
    ** (ArgumentError) errors were found at the given arguments:

  * 1st argument: invalid external representation of a term

        (cubdb 2.0.2) lib/cubdb/store/file.ex:156: CubDB.Store.CubDB.Store.File.raise_if_error/1
        (cubdb 2.0.2) lib/cubdb/btree.ex:64: CubDB.Btree.new/2
        (cubdb 2.0.2) lib/cubdb.ex:1232: CubDB.init/1
        (stdlib 4.1) gen_server.erl:851: :gen_server.init_it/2
        (stdlib 4.1) gen_server.erl:814: :gen_server.init_it/6
        (stdlib 4.1) proc_lib.erl:240: :proc_lib.init_p_do_apply/3

[lucaong]

It looks like CubDB successfully locates the latest good header, but then when it tries to read the root node that the header points to, it finds it's not a valid serialized representation.

This should not be caused by a crash: CubDB is append-only, and the transaction header is written last, so once a sane header is found, everything before it (including the root and the whole btree) must have been successfully written and committed. At a glance, it looks like file corruption of some sort. If you find a file that you can share with me (either here or privately) I will definitely investigate further. Meanwhile, if there is any reason to believe that the file system got corrupted, that could be it.

Regarding possible other causes, I assume that you did not downgrade OTP or erlang/Elixir to an earlier version, so I'd exclude a compatibility issue of the Erlang external format (also, I think it's pretty stable across releases).

If you have any other hint, do let me know. Thank you for reporting this!