Closed
Description
In 1.3.5, attempting to consume a message encoded with an unavailable codec (in my case, lz4) would give you a clear error message LZ4 decompression unsupported. However, in 1.4.1 I'm getting a far more confusing error:
File "/usr/lib/python2.7/site-packages/kafka_python-1.4.1-py2.7.egg/kafka/consumer/group.py", line 603, in poll
records = self._poll_once(remaining, max_records)
File "/usr/lib/python2.7/site-packages/kafka_python-1.4.1-py2.7.egg/kafka/consumer/group.py", line 652, in _poll_once
records, _ = self._fetcher.fetched_records(max_records)
File "/usr/lib/python2.7/site-packages/kafka_python-1.4.1-py2.7.egg/kafka/consumer/fetcher.py", line 337, in fetched_records
self._next_partition_records = self._parse_fetched_data(completion)
File "/usr/lib/python2.7/site-packages/kafka_python-1.4.1-py2.7.egg/kafka/consumer/fetcher.py", line 788, in _parse_fetched_data
unpacked = list(self._unpack_message_set(tp, records))
File "/usr/lib/python2.7/site-packages/kafka_python-1.4.1-py2.7.egg/kafka/consumer/fetcher.py", line 455, in _unpack_message_set
for record in batch:
File "/usr/lib/python2.7/site-packages/kafka_python-1.4.1-py2.7.egg/kafka/record/default_records.py", line 249, in __iter__
self._maybe_uncompress()
File "/usr/lib/python2.7/site-packages/kafka_python-1.4.1-py2.7.egg/kafka/record/default_records.py", line 165, in _maybe_uncompress
uncompressed = lz4_decode(data.tobytes())
TypeError: 'NoneType' object is not callable
I haven't looked too closely, but it looks like the new kafka.record codepath doesn't check has_lz4() before calling lz4_decode, causing the above.
Metadata
Metadata
Assignees
Labels
No labels