fluentd-sidecar-injector

fluentd-sidecar-injector is a webhook server for kubernetes admission webhook. This server inject fluentd or fluent-bit container as sidecar for specified Pod using mutation webhook. The feature is

• Automatically sidecar injection
• You can control injection using Pod's annotations
• You can change fluentd or fluent-bit docker image to be injected

Install

You can install this controller and webhook server using helm.

$ helm repo add h3poteto-stable https://h3poteto.github.io/charts/stable
$ helm install my-injector --namespace kube-system h3poteto-stable/fluentd-sidecar-injector

Please refer helm repository for parameters.

After install it, custom resources and controller will be installed.

$ kubectl get sidecarinjectors -n kube-system
NAME                  AGE
my-injector-fluentd   1m56s

$ kubectl get pods -n kube-system -l operator.h3poteto.dev=control-plane
NAME                                   READY   STATUS    RESTARTS   AGE
my-injector-manager-6d7f6bcd55-z5jcv   1/1     Running   0          2m17s

And it creates admission webhook for the sidecar injector.

$ kubectl get mutatingwebhookconfigurations
NAME                                           WEBHOOKS   AGE
sidecar-injector-webhook-my-injector-fluentd   1          5m15s

$ kubectl get svc -n kube-system -l sidecarinjectors.operator.h3poteto.dev=webhook-service
NAME                                   TYPE        CLUSTER-IP      EXTERNAL-IP   PORT(S)   AGE
sidecar-injector-my-injector-fluentd   ClusterIP   100.69.147.98   <none>        443/TCP   4m2s

$ kubectl get pods -n kube-system -l sidecarinjectors.operator.h3poteto.dev=webhook-pod
NAME                                           READY   STATUS    RESTARTS   AGE
my-injector-fluentd-handler-5969df9695-ftklp   1/1     Running   0          4m51s
my-injector-fluentd-handler-5969df9695-x5n5r   1/1     Running   0          4m51s

Usage

After you install this webhook server, fluentd sidecar containers are automatically injected, if you specify the annotation fluentd-sidecar-injector.h3poteto.dev/injection: 'enabled' to the pods.

For example:

apiVersion: apps/v1
kind: Deployment
metadata:
  name: nginx-test
  labels:
    app: nginx-test
spec:
  replicas: 1
  selector:
    matchLabels:
      app: nginx-test
  template:
    metadata:
      annotations:
        fluentd-sidecar-injector.h3poteto.dev/injection: 'enabled'
        fluentd-sidecar-injector.h3poteto.dev/application-log-dir: '/var/log/nginx'
        fluentd-sidecar-injector.h3poteto.dev/collector: 'fluentd'
      labels:
        app: nginx-test
    spec:
      containers:
        - name: nginx
          image: nginx:latest

FluentD is injected for this Pod.

$ kubectl get pod
NAME                          READY   STATUS    RESTARTS   AGE
nginx-test-6cbf4485f8-kq8ws   2/2     Running   0          9s

$ kubectl describe pod nginx-test-6cbf4485f8-kq8ws
Name:           nginx-test-6cbf4485f8-kq8ws
Namespace:      default
Containers:
  nginx:
    Container ID:   docker://ce74393381205786668a1fe2a4bc83ba058d380714b8a7ddca23966c8c7f0eb0
    Image:          nginx:latest
    Image ID:       docker-pullable://[email protected]:ad5552c786f128e389a0263104ae39f3d3c7895579d45ae716f528185b36bc6f
    Port:           <none>
    Host Port:      <none>
    State:          Running
      Started:      Fri, 14 Feb 2020 13:49:21 +0900
    Ready:          True
    Restart Count:  0
    Environment:    <none>
    Mounts:
      /var/log/nginx from fluentd-sidecar-injector-logs (rw)
      /var/run/secrets/kubernetes.io/serviceaccount from default-token-8rcns (ro)
  fluentd-sidecar:
    Container ID:   docker://49503c3836fa5ebc40c55db3717f16f21fbdbfaae8859a8ed8a366d04a2b6d9b
    Image:          ghcr.io/h3poteto/fluentd-forward:latest
    Image ID:       docker-pullable://ghcr.io/h3poteto/[email protected]:5d93af333ad9fefbfcb8013d20834fd89c2bbd3fe8b9b9bfa620ded29d7b3205
    Port:           <none>
    Host Port:      <none>
    State:          Running
      Started:      Fri, 14 Feb 2020 13:49:23 +0900
    Ready:          True
    Restart Count:  0
    Limits:
      memory:  1000Mi
    Requests:
      cpu:     100m
      memory:  200Mi
    Environment:
      AGGREGATOR_HOST:      127.0.0.1
      APPLICATION_LOG_DIR:  /var/log/nginx
      TAG_PREFIX:           prod
      TIME_KEY:             time
    Mounts:
      /var/log/nginx from fluentd-sidecar-injector-logs (rw)

Custom fluent.conf

If you need to use your own fluent.conf, use config-volume option. The following yaml has fluent-conf configmap. It will be mounted on /fluentd/etc/fluent/fluent.conf.

apiVersion: apps/v1
kind: Deployment
metadata:
  name: nginx-test
  labels:
    app: nginx-test
spec:
  replicas: 1
  selector:
    matchLabels:
      app: nginx-test
  template:
    metadata:
      annotations:
        fluentd-sidecar-injector.h3poteto.dev/injection: 'enabled'
        fluentd-sidecar-injector.h3poteto.dev/collector: 'fluentd'
        fluentd-sidecar-injector.h3poteto.dev/docker-image: 'fluent/fluentd:latest'
        fluentd-sidecar-injector.h3poteto.dev/application-log-dir: '/var/log/nginx'
        fluentd-sidecar-injector.h3poteto.dev/aggregator-host: 'fluentd.example.com'
        fluentd-sidecar-injector.h3poteto.dev/config-volume: 'fluent-conf'
      labels:
        app: nginx-test
    spec:
      containers:
        - name: nginx
          image: nginx:latest
    volumes:
      - name: fluent-conf
        configMap:
          name: fluent-conf
---
apiVersion: v1
kind: ConfigMap
metadata:
  name: fluent-conf
  labels:
    app: fluent-conf
data:
  fluent.conf: |-
    <source>
      @type tail
      path "#{ENV['APPLICATION_LOG_DIR']}/*.access.log"
      pos_file /var/tmp/application.log.pos
      tag "app.*"
      <parse>
        @type ltsv
      </parse>
    </source>

    <filter app.*>
      @type record_transformer
      <record>
        hostname "#{Socket.gethostname}"
      </record>
    </filter>

    <match app.*>
      @type forward

      <server>
        host "#{ENV['AGGREGATOR_HOST']}"
        port "#{ENV['AGGREGATOR_PORT']} || 24224"
      </server>
    </match>

Annotations

Please specify these annotations to your pods like this.

These annotations are used when collector is fluentd.

These annotations are used when collector is fluent-bit.

• fluentd-sidecar-injector.h3poteto.dev/injection specifies whether enable or disable this injector. Please specify enabled if you want to enable.
• fluentd-sidecar-injector.h3poteto.dev/docker-image specifies sidecar docker image. Default is ghcr.io/h3poteto/fluentd-forward:latest.
• fluentd-sidecar-injector.h3poteto.dev/collector specifies collector name which is fluentd or fluent-bit. Default is fluentd. Specified collector is injected you pods.
• fluentd-sidecar-injector.h3poteto.dev/aggregator-host is used in here. Default docker image forward received logs to another fluentd host. This parameter is required.
• fluentd-sidecar-injector.h3poteto.dev/aggregator-port is used in here. Default is 24224.
• fluentd-sidecar-injector.h3poteto.dev/application-log-dir specifies log directory where fluentd will watch. This directory is share between application container and sidecar fluentd container using volume mounts. This parameter is required.
• fluentd-sidecar-injector.h3poteto.dev/tag-prefix is prefix of received log's tag. It is used in here.
• fluentd-sidecar-injector.h3poteto.dev/config-volume can read your own fluent.conf. If you specify collector to fluent-bit, fluent-bit.conf is read.
• fluentd-sidecar-injector.h3poteto.dev/custom-env is an option that allows users to set their own values ​​in fluent.conf. Use with config-volume option.
• fluentd-sidecar-injector.h3poteto.dev/expose-port is an option that users can set any port to expose fluentd container.
• fluentd-sidecar-injector.h3poteto.dev/send-timeout is send timeout of fluentd configuration in here. Default is 60s.
• fluentd-sidecar-injector.h3poteto.dev/recover-wait is used in here. Default is 10s.
• fluentd-sidecar-injector.h3poteto.dev/hard-timeout is timeout of fluentd configuration in here. Default is 120s.
• fluentd-sidecar-injector.h3poteto.dev/time-key is fluentd configuration in here. Default is time.
• fluentd-sidecar-injector.h3poteto.dev/time-format is fluentd configuration in here. Default is %Y-%m-%dT%H:%M:%S%z.
• fluentd-sidecar-injector.h3poteto.dev/log-format is fluentd configuration in here. Default is json.
• fluentd-sidecar-injector.h3poteto.dev/refresh-interval is fluent-bit configuration in hrere. Default is 60 second.
• fluentd-sidecar-injector.h3poteto.dev/rotate-wait is fluent-bit configuration in here. Default is 5 second.

Fixed environment variables

The following values ​​will be set for each fluentd-sidecar. You can use this value in your fluent.conf with config-volume option.

You can find out more about the values on The Downward API.

Development

Please prepare a Kubernetes cluster to install this, and export KUBECONFIG.

$ export KUBECONFIG=$HOME/.kube/config

At first, build docker and push it.

$ docker build -t my-docker-registry/fluentd-sidecar-injector:experimental .
$ docker push my-docker-registry/fluentd-sidecar-injector:experimental

Next, install CRDs.

$ make install

Next, please run controller in local.

$ export WEBHOOK_CONTAINER_IMAGE=my-docker-registry/fluentd-sidecar-injector:experimental
$ export POD_NAMESPACE=my-namespace
$ make run

License

The package is available as open source under the terms of the MIT License.