import tempfile
import threading
import time
+import traceback
+
from apiclient import errors as apiclient_errors
from arvados._version import __version__
self._checkpointer.daemon = True
self._update_task_time = update_time # How many seconds wait between update runs
self._files_to_upload = FileUploadList(dry_run=dry_run)
+ self._upload_started = False
self.logger = logger
self.dry_run = dry_run
+ self._checkpoint_before_quit = True
if not self.use_cache and self.resume:
raise ArvPutArgumentConflict('resume cannot be True when use_cache is False')
# report initial progress.
self._update()
# Actual file upload
+ self._upload_started = True # Used by the update thread to start checkpointing
self._upload_files()
+ except (SystemExit, Exception) as e:
+ self._checkpoint_before_quit = False
+ # Log stack trace only when Ctrl-C isn't pressed (SIGINT)
+ # Note: We're expecting SystemExit instead of KeyboardInterrupt because
+ # we have a custom signal handler in place that raises SystemExit with
+ # the catched signal's code.
+ if not isinstance(e, SystemExit) or e.code != -2:
+ self.logger.warning("Abnormal termination:\n{}".format(traceback.format_exc(e)))
+ raise
finally:
if not self.dry_run:
# Stop the thread before doing anything else
self._stop_checkpointer.set()
self._checkpointer.join()
- # Commit all pending blocks & one last _update()
- self._local_collection.manifest_text()
- self._update(final=True)
- if save_collection:
- self.save_collection()
+ if self._checkpoint_before_quit:
+ # Commit all pending blocks & one last _update()
+ self._local_collection.manifest_text()
+ self._update(final=True)
+ if save_collection:
+ self.save_collection()
if self.use_cache:
self._cache_file.close()
Periodically called support task. File uploading is
asynchronous so we poll status from the collection.
"""
- while not self._stop_checkpointer.wait(self._update_task_time):
+ while not self._stop_checkpointer.wait(1 if not self._upload_started else self._update_task_time):
self._update()
def _update(self, final=False):
"""
Update cached manifest text and report progress.
"""
- with self._collection_lock:
- self.bytes_written = self._collection_size(self._local_collection)
+ if self._upload_started:
+ with self._collection_lock:
+ self.bytes_written = self._collection_size(self._local_collection)
+ if self.use_cache:
+ if final:
+ manifest = self._local_collection.manifest_text()
+ else:
+ # Get the manifest text without comitting pending blocks
+ manifest = self._local_collection.manifest_text(strip=False,
+ normalize=False,
+ only_committed=True)
+ # Update cache
+ with self._state_lock:
+ self._state['manifest'] = manifest
if self.use_cache:
- if final:
- manifest = self._local_collection.manifest_text()
- else:
- # Get the manifest text without comitting pending blocks
- manifest = self._local_collection.manifest_text(strip=False,
- normalize=False,
- only_committed=True)
- # Update cache
- with self._state_lock:
- self._state['manifest'] = manifest
- if self.use_cache:
- self._save_state()
+ self._save_state()
+ else:
+ self.bytes_written = self.bytes_skipped
# Call the reporter, if any
self.report_progress()
return self._my_collection().manifest_locator()
def portable_data_hash(self):
- return self._my_collection().portable_data_hash()
+ pdh = self._my_collection().portable_data_hash()
+ m = self._my_collection().stripped_manifest()
+ local_pdh = hashlib.md5(m).hexdigest() + '+' + str(len(m))
+ if pdh != local_pdh:
+ logger.warning("\n".join([
+ "arv-put: API server provided PDH differs from local manifest.",
+ " This should not happen; showing API server version."]))
+ return pdh
def manifest_text(self, stream_name=".", strip=False, normalize=False):
return self._my_collection().manifest_text(stream_name, strip, normalize)
global api_client
logger = logging.getLogger('arvados.arv_put')
+ logger.setLevel(logging.INFO)
args = parse_arguments(arguments)
status = 0
if api_client is None:
else:
reporter = None
+ # If this is used by a human, and there's at least one directory to be
+ # uploaded, the expected bytes calculation can take a moment.
+ if args.progress and any([os.path.isdir(f) for f in args.paths]):
+ logger.info("Calculating upload size, this could take some time...")
bytes_expected = expected_bytes_for(args.paths)
try: