Enterprise Kubernetes, delivered

Tectonic ships with CoreOS's signature automated operations, runs multi-cloud, and is the fastest, most secure path to Kubernetes.

Customizing Fluentd

The fluentd-configmap.yaml provided has been designed to be easily customizable. Generally you'll want to avoid modifying anything other than the fluentd.conf and output.conf sections of the configmap.

Customizing log destination

The customizing log destination document explains how to configure where logs are sent.

Add custom parsing/filtering logic

To add additional filters or parsers, add them to the extra.conf section in the fluentd-configmap.yaml. The extra.conf already has a very brief example of how to add an extra field to log entries, and a more detailed example is shown below.

For details on Fluentd post-processing, check out the Fluentd filters and parsers documents.

Targeting a specific application's logs

Fluentd routes event based on tags. Events flowing through Fluentd can be routed based on the value of the tag using <match> and <filter> directives. The configuration tags events using the following conventions:

  • systemd logs have a tag systemd.<unit-name-here>
  • Kubernetes container logs have a tag kube.<namespace-name>.<container-name>
  • Kubernetes API audit logs have a tag kube-apiserver-audit

The existing configuration already does additional post processing based on some of these tags.

For example, the host's kubelet.service log's are parsed by matching on the tag systemd.kubelet, and we do the same for parsing the Docker engine's logs using the tag systemd.docker. These filters set their key_name parameter to MESSAGE which is the actual field name for the log message when it originates from journald.

Similarly, we parse the logs of the kube-apiserver, kube-scheduler, and other controller components by performing a wildcard match on the tag: kube.kube-system.**. This filter set its key_name parameter to log, which is the field for log messages originating from Docker.

Example: Parsing the guestbook apache logs

The following configuration will parse the frontend component's logs from the guestbook example app deployed in the "Deploy your second app" tutorial. To use it, copy and paste the snippet below fluentd-configmap.yaml's extra.conf section (make sure you indent to the correct level).

<filter kube.default.php-redis>
  @type parser
  # Fluentd provides a few built-in formats for popular and common formats such as "apache" and "json".
  format apache2
  key_name log
  # Retain the original "log" field after parsing out the data.
  reserve_data true

  # The access logs and error logs are interleaved with each other and have
  # different formats, so ignore parse errors, as they're expected
  suppress_parse_error_log true
</filter>

<filter kube.default.php-redis>
  @type parser
  format apache_error
  key_name log
  reserve_data true

  # The access logs and error logs are interleaved with each other and have
  # different formats, so ignore parse errors, as they're expected
  suppress_parse_error_log true
</filter>

Once you've updated your config, you will need to delete and recreate your fluentd pods in order for the configuration to take effect:

$ kubectl delete pods --namespace logging -l app=fluentd

Once your pods have restarted, any new logs being parsed should be using the new configuration.