Writing logs to Opensearch¶
Added in provider version 1.5.0 Available only with Airflow>=3.0
Airflow can be configured to read task logs from Opensearch and optionally write logs to stdout in standard or json format. These logs can later be collected and forwarded to the cluster using tools like fluentd, logstash or others.
Airflow also supports writing logs to OpenSearch directly without requiring additional software like fluentd or logstash. To enable this feature, set write_to_os and json_format to True and write_stdout to False in airflow.cfg.
You can choose to have all task logs from workers output to the highest parent level process, instead of the standard file locations. This allows for some additional flexibility in container environments like Kubernetes, where container stdout is already being logged to the host nodes. From there a log shipping tool can be used to forward them along to Opensearch. To use this feature, set the write_stdout option in airflow.cfg.
You can also choose to have the logs output in a JSON format, using the json_format option. Airflow uses the standard Python logging module and JSON fields are directly extracted from the LogRecord object. To use this feature, set the json_fields option in airflow.cfg. Add the fields to the comma-delimited string that you want collected for the logs. These fields are from the LogRecord object in the logging module. Documentation on different attributes can be found here.
First, to use the handler, airflow.cfg must be configured as follows:
[logging]
remote_logging = True
[opensearch]
host = <host>
port = <port>
username = <username>
password = <password>
To output task logs to stdout in JSON format, the following config could be used:
[logging]
remote_logging = True
[opensearch]
write_stdout = True
json_format = True
To output task logs to OpenSearch directly, the following config could be used: (set delete_local_logs to True if you do not want to retain a local copy of the task log)
[logging]
remote_logging = True
delete_local_logs = False
[opensearch]
host = <host>
port = <port>
username = <username>
password = <password>
write_stdout = False
json_format = True
write_to_os = True
target_index = [name of the index to store logs]
Writing logs to Opensearch over TLS¶
To add custom configurations to Opensearch (e.g. turning on ssl_verify, adding a custom self-signed
cert, etc.) use the opensearch_configs setting in your airflow.cfg
Note that these configurations also apply when you enable writing logs to OpenSearch directly.
[logging]
remote_logging = True
[opensearch_configs]
use_ssl = True
verify_certs = True
ssl_assert_hostname = True
ca_certs=/path/to/CA_certs