Merge pull request #2926 from lbryio/blocking-file-io

make file i/o in stream and blob creation and reads non-blocking

lbry/blob/blob_file.py

@@ -110,7 +110,7 @@ class AbstractBlob:
                 if reader in self.readers:
                     self.readers.remove(reader)
 
-    def _write_blob(self, blob_bytes: bytes):
+    def _write_blob(self, blob_bytes: bytes) -> asyncio.Task:
         raise NotImplementedError()
 
     def set_length(self, length) -> None:
@@ -198,11 +198,17 @@ class AbstractBlob:
     def save_verified_blob(self, verified_bytes: bytes):
         if self.verified.is_set():
             return
-        if self.is_writeable():
-            self._write_blob(verified_bytes)
+
+        def update_events(_):
             self.verified.set()
+            self.writing.clear()
+
+        if self.is_writeable():
+            self.writing.set()
+            task = self._write_blob(verified_bytes)
+            task.add_done_callback(update_events)
             if self.blob_completed_callback:
-                self.blob_completed_callback(self)
+                task.add_done_callback(lambda _: self.blob_completed_callback(self))
 
     def get_blob_writer(self, peer_address: typing.Optional[str] = None,
                         peer_port: typing.Optional[int] = None) -> HashBlobWriter:
@@ -261,9 +267,11 @@ class BlobBuffer(AbstractBlob):
             self.verified.clear()
 
     def _write_blob(self, blob_bytes: bytes):
+        async def write():
             if self._verified_bytes:
                 raise OSError("already have bytes for blob")
             self._verified_bytes = BytesIO(blob_bytes)
+        return self.loop.create_task(write())
 
     def delete(self):
         if self._verified_bytes:
@@ -319,9 +327,15 @@ class BlobFile(AbstractBlob):
             handle.close()
 
     def _write_blob(self, blob_bytes: bytes):
+        def _write_blob():
             with open(self.file_path, 'wb') as f:
                 f.write(blob_bytes)
 
+        async def write_blob():
+            await self.loop.run_in_executor(None, _write_blob)
+
+        return self.loop.create_task(write_blob())
+
     def delete(self):
         if os.path.isfile(self.file_path):
             os.remove(self.file_path)

lbry/blob_exchange/client.py

@@ -152,6 +152,8 @@ class BlobExchangeClientProtocol(asyncio.Protocol):
             log.debug(msg)
             msg = f"downloaded {self.blob.blob_hash[:8]} from {self.peer_address}:{self.peer_port}"
             await asyncio.wait_for(self.writer.finished, self.peer_timeout, loop=self.loop)
+            # wait for the io to finish
+            await self.blob.verified.wait()
             log.info("%s at %fMB/s", msg,
                      round((float(self._blob_bytes_received) /
                             float(time.perf_counter() - start_time)) / 1000000.0, 2))

lbry/stream/descriptor.py

@@ -44,16 +44,23 @@ def random_iv_generator() -> typing.Generator[bytes, None, None]:
         yield os.urandom(AES.block_size // 8)
 
 
-def file_reader(file_path: str):
+def read_bytes(file_path: str, offset: int, to_read: int):
+    with open(file_path, 'rb') as f:
+        f.seek(offset)
+        return f.read(to_read)
+
+
+async def file_reader(file_path: str):
     length = int(os.stat(file_path).st_size)
     offset = 0
 
-    with open(file_path, 'rb') as stream_file:
     while offset < length:
         bytes_to_read = min((length - offset), MAX_BLOB_SIZE - 1)
         if not bytes_to_read:
             break
-            blob_bytes = stream_file.read(bytes_to_read)
+        blob_bytes = await asyncio.get_event_loop().run_in_executor(
+            None, read_bytes, file_path, offset, bytes_to_read
+        )
         yield blob_bytes
         offset += bytes_to_read
 
@@ -245,7 +252,7 @@ class StreamDescriptor:
         iv_generator = iv_generator or random_iv_generator()
         key = key or os.urandom(AES.block_size // 8)
         blob_num = -1
-        for blob_bytes in file_reader(file_path):
+        async for blob_bytes in file_reader(file_path):
             blob_num += 1
             blob_info = await BlobFile.create_from_unencrypted(
                 loop, blob_dir, key, next(iv_generator), blob_bytes, blob_num, blob_completed_callback

lbry/stream/managed_stream.py

@@ -343,7 +343,8 @@ class ManagedStream:
                 self.streaming.clear()
 
     @staticmethod
-    def _write_decrypted_blob(handle: typing.IO, data: bytes):
+    def _write_decrypted_blob(output_path: str, data: bytes):
+        with open(output_path, 'ab') as handle:
             handle.write(data)
             handle.flush()
 
@@ -355,10 +356,10 @@ class ManagedStream:
         self.finished_writing.clear()
         self.started_writing.clear()
         try:
-            with open(output_path, 'wb') as file_write_handle:
+            open(output_path, 'wb').close()
             async for blob_info, decrypted in self._aiter_read_stream(connection_id=self.SAVING_ID):
                 log.info("write blob %i/%i", blob_info.blob_num + 1, len(self.descriptor.blobs) - 1)
-                    await self.loop.run_in_executor(None, self._write_decrypted_blob, file_write_handle, decrypted)
+                await self.loop.run_in_executor(None, self._write_decrypted_blob, output_path, decrypted)
                 if not self.started_writing.is_set():
                     self.started_writing.set()
             await self.update_status(ManagedStream.STATUS_FINISHED)

tests/unit/blob/test_blob_manager.py

@@ -16,7 +16,7 @@ class TestBlobManager(AsyncioTestCase):
         self.blob_manager = BlobManager(self.loop, tmp_dir, self.storage, self.config)
         await self.storage.open()
 
-    async def test_memory_blobs_arent_verifie_but_real_ones_are(self):
+    async def test_memory_blobs_arent_verified_but_real_ones_are(self):
         for save_blobs in (False, True):
             await self.setup_blob_manager(save_blobs=save_blobs)
             # add a blob file
@@ -24,6 +24,7 @@ class TestBlobManager(AsyncioTestCase):
             blob_bytes = b'1' * ((2 * 2 ** 20) - 1)
             blob = self.blob_manager.get_blob(blob_hash, len(blob_bytes))
             blob.save_verified_blob(blob_bytes)
+            await blob.verified.wait()
             self.assertTrue(blob.get_is_verified())
             self.blob_manager.blob_completed(blob)
             self.assertEqual(self.blob_manager.is_blob_verified(blob_hash), save_blobs)

tests/unit/blob_exchange/test_transfer_blob.py

@@ -130,10 +130,14 @@ class TestBlobExchange(BlobExchangeTestBase):
         write_blob = blob._write_blob
         write_called_count = 0
 
-        def wrap_write_blob(blob_bytes):
+        async def _wrap_write_blob(blob_bytes):
             nonlocal write_called_count
             write_called_count += 1
-            write_blob(blob_bytes)
+            await write_blob(blob_bytes)
+
+        def wrap_write_blob(blob_bytes):
+            return asyncio.create_task(_wrap_write_blob(blob_bytes))
+
         blob._write_blob = wrap_write_blob
 
         writer1 = blob.get_blob_writer(peer_port=1)
@@ -166,6 +170,7 @@ class TestBlobExchange(BlobExchangeTestBase):
 
         self.assertDictEqual({1: mock_blob_bytes, 2: mock_blob_bytes}, results)
         self.assertEqual(1, write_called_count)
+        await blob.verified.wait()
         self.assertTrue(blob.get_is_verified())
         self.assertDictEqual({}, blob.writers)