18941: Add --threads option to arv-get

[arvados.git] / sdk / python / arvados / arvfile.py

diff --git a/sdk/python/arvados/arvfile.py b/sdk/python/arvados/arvfile.py
index b21ebd331769109ca0cca3b23e2c45f17ed2c9ed..fbf593d02619810b1647f78520217ac6a9b4bef7 100644 (file)
--- a/sdk/python/arvados/arvfile.py
+++ b/sdk/python/arvados/arvfile.py
@@ -479,9 +479,9 @@ class _BlockManager(object):
     """
 
     DEFAULT_PUT_THREADS = 2
-    DEFAULT_GET_THREADS = 4
+    DEFAULT_GET_THREADS = 2
 
-    def __init__(self, keep, copies=None, put_threads=None, num_retries=None, storage_classes_func=None):
+    def __init__(self, keep, copies=None, put_threads=None, num_retries=None, storage_classes_func=None, get_threads=None):
         """keep: KeepClient object to use"""
         self._keep = keep
         self._bufferblocks = collections.OrderedDict()
@@ -492,7 +492,7 @@ class _BlockManager(object):
         self.lock = threading.Lock()
         self.prefetch_enabled = True
         self.num_put_threads = put_threads or _BlockManager.DEFAULT_PUT_THREADS
-        self.num_get_threads = _BlockManager.DEFAULT_GET_THREADS
+        self.num_get_threads = get_threads or _BlockManager.DEFAULT_GET_THREADS
         self.copies = copies
         self.storage_classes = storage_classes_func or (lambda: [])
         self._pending_write_size = 0
@@ -1103,7 +1103,7 @@ class ArvadosFile(object):
             if size == 0 or offset >= self.size():
                 return b''
             readsegs = locators_and_ranges(self._segments, offset, size)
-            prefetch = locators_and_ranges(self._segments, offset + size, config.KEEP_BLOCK_SIZE*4, limit=32)
+            prefetch = locators_and_ranges(self._segments, offset + size, config.KEEP_BLOCK_SIZE * self.parent._my_block_manager().num_get_threads, limit=32)
 
         locs = set()
         data = []
@@ -1121,10 +1121,7 @@ class ArvadosFile(object):
                 self.parent._my_block_manager().block_prefetch(lr.locator)
                 locs.add(lr.locator)
 
-        if len(data) == 1:
-            return data[0]
-        else:
-            return b''.join(data)
+        return b''.join(data)
 
     @must_be_writable
     @synchronized