def run(self):
# batch stream must be executed serially, and reinit.
# TODO:0: remove lock to bss
rs_header = None
with PutBatchTask._partition_lock[self.tag]: # tag includes partition info in tag generation
L.trace(f"do_store start for tag={self.tag}, partition_id={self.partition._id}")
bss = _BatchStreamStatus.get_or_create(self.tag)
try:
broker = TransferService.get_or_create_broker(self.tag, write_signals=1)
iter_wait = 0
iter_timeout = int(CoreConfKeys.EGGROLL_CORE_FIFOBROKER_ITER_TIMEOUT_SEC.get())
batch = None
batch_get_interval = 0.1
with create_adapter(self.partition) as db, db.new_batch() as wb:
#for batch in broker:
while not broker.is_closable():
try:
batch = broker.get(block=True, timeout=batch_get_interval)
except queue.Empty as e:
iter_wait += batch_get_interval
if iter_wait > iter_timeout:
raise TimeoutError(f'timeout in PutBatchTask.run. tag={self.tag}, iter_timeout={iter_timeout}, iter_wait={iter_wait}')
else:
continue
except BrokerClosed as e:
continue
iter_wait = 0
rs_header = ErRollSiteHeader.from_proto_string(batch.header.ext)
batch_pairs = 0
if batch.data:
bin_data = ArrayByteBuffer(batch.data)
reader = PairBinReader(pair_buffer=bin_data, data=batch.data)
for k_bytes, v_bytes in reader.read_all():
wb.put(k_bytes, v_bytes)
batch_pairs += 1
bss.count_batch(rs_header, batch_pairs)
# TODO:0
bss._data_type = rs_header._data_type
if rs_header._stage == FINISH_STATUS:
bss.set_done(rs_header) # starting from 0
bss.check_finish()
# TransferService.remove_broker(tag) will be called in get_status phrase finished or exception got
except Exception as e:
L.exception(f'_run_put_batch error, tag={self.tag}, '
f'rs_key={rs_header.get_rs_key() if rs_header is not None else None}, rs_header={rs_header}')
raise e
finally:
TransferService.remove_broker(self.tag)
def bin_batch_to_pair(input_iter):
L.debug(f"bin_batch_to_pair start")
total_written = 0
for batch in input_iter:
L.debug(f"bin_batch_to_pair batch start size:{len(batch)}")
try:
bin_data = ArrayByteBuffer(batch)
reader = PairBinReader(pair_buffer=bin_data)
for k_bytes, v_bytes in reader.read_all():
yield k_bytes, v_bytes
total_written += 1
except IndexError as e:
L.exception(f"error bin bath format:{e}")
L.debug(f"bin_batch_to_pair batch end count:{total_written}")
L.debug(f"bin_batch_to_pair total_written count:{total_written}")
def bin_batch_to_pair(input_iter):
L.trace(f"bin_batch_to_pair start")
write_count = 0
for batch in input_iter:
L.trace(f"bin_batch_to_pair: cur batch size={len(batch)}")
try:
bin_data = ArrayByteBuffer(batch)
reader = PairBinReader(pair_buffer=bin_data, data=batch)
for k_bytes, v_bytes in reader.read_all():
yield k_bytes, v_bytes
write_count += 1
except IndexError as e:
L.exception(f"error bin bath format: {e}")
L.trace(f"bin_batch_to_pair batch ends. total write count={write_count}")
L.trace(f"bin_batch_to_pair total_written count={write_count}")
def test_pair_bin(self):
bs = bytearray(32)
buf = ArrayByteBuffer(bs)
writer = PairBinWriter(buf)
for i in range(10):
try:
writer.write(str(i).encode(), str(i).encode())
except IndexError as e:
print(buf.read_bytes(buf.get_offset(), 0))
buf.set_offset(0)
writer = PairBinWriter(buf)
writer.write(str(i).encode(), str(i).encode())
buf.set_offset(0)
reader = PairBinReader(buf)
print("last")
print(list(reader.read_all()))
class FileIterator(PairIterator):
def __init__(self, file):
file.seek(0)
self.reader = PairBinReader(FileByteBuffer(file))
def close(self):
pass
def __iter__(self):
return self.reader.read_all()
class MmapIterator(PairIterator):
def __init__(self, file):
self.mm = mmap.mmap(file.fileno(), 0)
self.mm.seek(0)
self.reader = PairBinReader(ArrayByteBuffer(self.mm))
def close(self):
pass
def __iter__(self):
return self.reader.read_all()
def test_pair_bin_no_abb(self):
bs = bytearray(32)
buf = ArrayByteBuffer(bs)
writer = PairBinWriter(pair_buffer=buf, data=bs)
for i in range(10):
try:
writer.write(str(i).encode(), str(i).encode())
except IndexError as e:
writer.set_offset(0)
writer = PairBinWriter(pair_buffer=buf, data=bs)
writer.write(str(i).encode(), str(i).encode())
pbr = PairBinReader(pair_buffer=buf, data=writer.get_data())
print(pbr.read_bytes(writer.get_offset(), 0))
writer.set_offset(0)
reader = PairBinReader(pair_buffer=buf, data=bs)
print("last")
print(list(reader.read_all()))
def __init__(self, file):
self.mm = mmap.mmap(file.fileno(), 0)
self.mm.seek(0)
self.reader = PairBinReader(ArrayByteBuffer(self.mm))
def __init__(self, file):
file.seek(0)
self.reader = PairBinReader(FileByteBuffer(file))