python-logstash-async

• Breaking change: drop support for 3.8, 3.9 and 3.10.
  • Catch errors on TCP/UDP socket closing to prevent requeuing of events (#98, #100).
  • Add new setting to configure batch sizes for Beats transport (#93).
  • Support URL path in HttpTransport (#103, Seyyed Mohammad Borghei).
  • LogstashFormatter: Move top_level_field_set creation to init to ease setting FORMATTER_LOGSTASH_MESSAGE_FIELD_LIST after importing the formatter class (#96, #97, feliixx).

• Add ECS (Elastic Common Schema) formatter support (#91, Andrii Lahuta).
  • On closing UDP/TCP socket, make sure it is reset (#89, #92).
  • Ensure to not check for socket write buffer if disabled.

• Implement get_non_flushed_event_count() for MemoryCache (#88).

• Wait for empty socket write buffer only if "fnctl" is available (#82).
  • Wait at most 30s for empty socket write buffer (#83).
  • Read pending events from database on startup for immediate flushing (#85).

• Add a wait until all data is sent, before closing socket on TCP transport (#81, Alisher Nazarkhanov).
  • Explicitly shutdown the socket connection in Udp/TcpTransport to prevent peer connection errors.
  • Implement optional database VACUUM on shutdown (#80).

• Load certificate chain only if a certificate was specified (#79).
  • Handle network errors raised by "requests" in HttpTransport (#75).

• Handle database disk errors (#72, Peter Mazarovich).
  • Use Python's PriorityQueue to reduce memory consumption (#73, Peter Mazarovich).

• Update link to docs for rate limit string notation (#70, Garrett Hyde).
  • Log connection and network errors as warnings (#71).
  • Fix support for Django 4.0 by using Request.build_absolute_uri().
  • Use datetime as column type for "entry_date" in database.

• Consider set also as sequence type when formatting events.
  • Improve formatter performance by converting simple types first (#64, Johann Schmitz).
  • Migrate from ssl.wrap_socket() to SSLContext.wrap_socket().

• Add SynchronousLogstashHandler: operates like the AsynchronousLogstashHandler but synchronously and without any event queueing and retrying (#59, #60, Sascha Pfeiffer).

• Fix missing message field in events (#58).

• Remove "six" dependency.
• Add "HTTP Input" transport (#56, Jürgen Löhel).
• Skip logging stacktraces for common network errors (#55, #56, Walter Macomber).
• Extend FORMATTER_RECORD_FIELD_SKIP_LIST to filter all record fields (#52).
• Docs: Fix error in logging.config.fileConfig code example (#54, Guy).

• Set python_requires to Python >=2.7 or Python >= 3.5 for smooth upgrade to upcoming Python3 only. This way Python2 only users will stay at this release.

• Fix socket timeout setting ignored for filebeat (#50, Koert van der Veer).

• Add "@metadata" to the generated event in Formatter, useful for common beats input configuration in Logstash (#49, Sudheer Satyanarayana).
• Don't convert text to bytes in Formatter (fix #45) (#46, Sergey Trofimov).

• Fix accessing request's META attribute in DjangoLogstashFormatter

• Handle DisallowedHost exceptions in DjangoLogstashFormatter to not trigger an exception while formatting.

• When sending all events, stop on errors. Otherwise retrying happens forever on non-recoverable errors.

• Flush all pending events on shutdown

• Docs: Explain the differences to FileBeat (#44)
  • Close TCP socket on connect timeouts
  • Add Flask formatter

• Fix broken transport instantiation if callable is used (#42)
  • Fix tcp input codec in documentation

• Add new constant QUEUED_EVENTS_BATCH_SIZE to limit events sent once (related to #31)
  • Add "Beats" transport (logstash-input-beats, related to #31)
  • Add "timeout" argument to transport classes for more flexible setting
  • Docs: better document the shared database between handlers (#30)
  • Perform tests also against Python 3.7
  • Docs: explain Formatter extra dict in more detail (#23)
  • Docs: minor improvements

• Handle possible non-bytes result in Formatter (#18, rmihael)

• Implement rate limiting of LogProcessingWorker error messages
  • Add a constant to configure the timeout for the Sqlite database

• Call flush on worker thread only if it is running (prevent errors when application shuts down before anything was logged)
• Rewrite utils.ichunked to not raise StopIteration (for future Python 3.7 compability)

For details see https://pypi.python.org/pypi/python-logstash-async/1.3.1

• Added in-memory cache back (#12, loganasherjones)
• Added support for TTL of messages (loganasherjones)
• Minor Python3 compability fixes
• Implement AsynchronousLogstashHandler::flush method to manually flushed queued events (#15, Vladislav Pakhomov)
• Ease modification of constants from the calling application (#14)
• Add Formatter parameter 'ensure_ascii' to workaround potential encoding errors on some setups (#9, Sergei Lobastov)

For details see https://pypi.python.org/pypi/python-logstash-async/1.3.0

• Require path setting to the Sqlite database (the previous default value ":memory:" doesn't help, see #5)
  • Fix a Python3 incompatibility (#3)
  • Fix a Django 1.8 / local Django development server incompatibility (#3)

For details see https://pypi.python.org/pypi/python-logstash-async/1.2.0

• Improve handling on locked Sqlite database on high event load

For details see https://pypi.python.org/pypi/python-logstash-async/1.1.1

• Add documentation (built by Sphinx, hosted at readthedocs.org)
• Fix installing of “six” dependency on package install

For details see https://pypi.python.org/pypi/python-logstash-async/1.1.0

Bugfix for #2.

For details see https://pypi.python.org/pypi/python-logstash-async/1.0.1

For details see https://pypi.python.org/pypi/python-logstash-async/1.0.0