Zack Williams | 9e8efd3 | 2018-10-17 15:01:13 -0700 | [diff] [blame^] | 1 | KafkaLogHandler |
| 2 | =============== |
| 3 | |
| 4 | Provides a python ``logging`` compatible handler for producing messages to a |
| 5 | Kafka message bus. |
| 6 | |
| 7 | Depends on the confluent_kafka module to connect to Kafka |
| 8 | |
| 9 | Designed to support structured logging, and serializes log data as JSON when |
| 10 | published as a Kafka message. |
| 11 | |
| 12 | Usage |
| 13 | ===== |
| 14 | |
| 15 | **Example:** |
| 16 | |
| 17 | :: |
| 18 | |
| 19 | import logger |
| 20 | |
| 21 | from kafkaloghandler import KafkaLogHandler |
| 22 | |
| 23 | log = logging.getLogger() |
| 24 | |
| 25 | klh = KafkaLogHandler(bootstrap_servers=["test-kafka:9092"], topic="testtopic") |
| 26 | |
| 27 | log.addHandler(klh) |
| 28 | |
| 29 | data={'example':'structured data'} |
| 30 | |
| 31 | log.info('message to send to kafka', data=data) |
| 32 | |
| 33 | |
| 34 | **Parameters that can be provided to KafkaLogHandler:** |
| 35 | |
| 36 | *bootstrap_servers* |
| 37 | List of Kafka bootstrap servers to connect to. See confluent_kafka docs. |
| 38 | |
| 39 | **default:** ``["localhost:9092"]`` |
| 40 | |
| 41 | *timeout* |
| 42 | Timeout in seconds for flushing producer queue. See confluent_kafka docs. |
| 43 | |
| 44 | **default:** ``10.0`` |
| 45 | |
| 46 | *topic* |
| 47 | String that sets the topic in Kafka. |
| 48 | |
| 49 | **default:** ``"kafkaloghandler"`` |
| 50 | |
| 51 | *key* |
| 52 | String that sets the default key in Kafka, can be used for summarization within Kafka. |
| 53 | |
| 54 | NOTE: This default key can be overridden on a per-message basis by passing a |
| 55 | dict to the logger with ``{"key": "new_key_for_this_message"}`` in it. |
| 56 | |
| 57 | **default:** ``"klh"`` |
| 58 | |
| 59 | *flatten* |
| 60 | Flattens nested dictionary keys passed as structured logging into the parent |
| 61 | dictionary layer, up to a certain depth. |
| 62 | |
| 63 | This is useful when logging to external systems that don't have good support |
| 64 | for hierarchical data. |
| 65 | |
| 66 | Example: ``{'a': {'b': 'c'}}`` would be flattened to ``{'a.b': 'c'}`` |
| 67 | |
| 68 | If the depth is exceeded, any remaining deeper dict will be added to the |
| 69 | output under the flattened key. |
| 70 | |
| 71 | Set to ``0`` to turn off flattening. |
| 72 | |
| 73 | **default:** ``5`` |
| 74 | |
| 75 | *separator* |
| 76 | Separator used between keys when flattening. |
| 77 | |
| 78 | **default:** ``.`` |
| 79 | |
| 80 | *blacklist* |
| 81 | List of top-level keys to discard from structured logs when outputting JSON. |
| 82 | |
| 83 | **default:** ``['_logger']`` |
| 84 | |
| 85 | |
| 86 | Tests |
| 87 | ===== |
| 88 | |
| 89 | Unit tests can be run with: |
| 90 | |
| 91 | nose2 --verbose --coverage-report term |