Kubernetes Logging with Fluentd
Kubernetes provides two logging end-points for applications and cluster logs: Stackdriver Logging for use with Google Cloud Platform and Elasticsearch. Behind the scenes there is a logging agent that take cares of log collection, parsing and distribution: Fluentd.
The following document focus on how to deploy Fluentd in Kubernetes and extend the possibilities to have different destinations for your logs.
Table of Contents
The following document assumes that you have a Kubernetes cluster running or at least a local (single) node that can be used for testing purposes.
Before to get started, make sure you understand or have a basic idea about the following concepts from Kubernetes:
A node is a worker machine in Kubernetes, previously known as a minion. A node may be a VM or physical machine, depending on the cluster. Each node has the services necessary to run pods and is managed by the master components…
A pod (as in a pod of whales or pea pod) is a group of one or more containers (such as Docker containers), the shared storage for those containers, and options about how to run the containers. Pods are always co-located and co-scheduled, and run in a shared context…
A DaemonSet ensures that all (or some) nodes run a copy of a pod. As nodes are added to the cluster, pods are added to them. As nodes are removed from the cluster, those pods are garbage collected. Deleting a DaemonSet will clean up the pods it created…
Since applications runs in Pods and multiple Pods might exists across multiple nodes, we need a specific Fluentd-Pod that takes care of log collection on each node: Fluentd DaemonSet.
Fluentd is flexible enough and have the proper plugins to distribute logs to different third party applications like databases or cloud services, so the principal question is to know: where the logs will be stored ?. Once we got that question answered, we can move forward configuring our DaemonSet.
The below steps will focus on sending the logs to a Elasticsearch Pod.
Get Fluentd DaemonSet sources
We have created a Fluentd DaemonSet that have the proper rules and container image ready to get started:
Please grab a copy of the repository from the command line using GIT:
$ git clone https://github.com/fluent/fluentd-kubernetes-daemonset
The cloned repository contains the configuration that allows to deploy Fluentd as a DaemonSet, the Docker container image distributed on the repository also comes pre-configured so Fluentd can gather all logs from the Kubernetes node environment and also it appends the proper metadata to the logs.
By default, this configuration will send all incoming logs to Elasticsearch, but you can customize for any outputs as well.
Logging to Elasticsearch
From the fluentd-kubernetes-daemonset/ directory, find the Yaml configuration file:
As an example let’s see a part of the file content:
apiVersion: extensions/v1beta1 kind: DaemonSet metadata: name: fluentd namespace: kube-system ... spec: ... spec: containers: - name: fluentd image: quay.io/fluent/fluentd-kubernetes-daemonset env: - name: FLUENT_ELASTICSEARCH_HOST value: "elasticsearch-logging" - name: FLUENT_ELASTICSEARCH_PORT value: "9200" ...
The Yaml file have two relevant environment variables that are used by Fluentd when the container starts:
|FLUENT_ELASTICSEARCH_HOST||Specify the host name or IP address.||elasticsearch-logging|
|FLUENT_ELASTICSEARCH_PORT||Elasticsearch TCP port||9200|
Any relevant change needs to be done to the Yaml file before to deploy it. Using the default values assumes that at least an Elasticsearch Pod elasticsearch-logging exists in the cluster.
If this article is incorrect or outdated, or omits critical information, please let us know. Fluentd is a open source project under Cloud Native Computing Foundation (CNCF), originally invented by Treasure Data, Inc. All components are available under the Apache 2 License.