kubernetes/cluster/addons/fluentd-gcp/README.md

# Stackdriver Logging Agent
==============

Stackdriver Logging Agent is a DaemonSet which spawns a pod on each node
that reads logs, generated by kubelet, container runtime and containers
and sends them to the Stackdriver. When logs are exported to the Stackdriver,
they can be searched, viewed, and analyzed.

Learn more at: https://kubernetes.io/docs/tasks/debug-application-cluster/logging-stackdriver

## Troubleshooting

In Kubernetes clusters in version 1.10.0 or later, fluentd-gcp DaemonSet can be
manually scaled. This is useful e.g. when applications running in the cluster
are sending a large volume of logs (i.e. over 100kB/s), causing fluentd-gcp to
fail with OutOfMemory errors. Conversely, if the applications aren't generating
a lot of logs, it may be useful to reduce the amount of resources consumed by
fluentd-gcp, making these resources available to other applications. To learn
more about Kubernetes resource requests and limits, see the official
documentation ([CPU][cpu], [memory][memory]). The amount of resources requested
by fluentd-gcp on every node in the cluster can be fetched by running following
command:

```
$ kubectl get ds -n kube-system -l k8s-app=fluentd-gcp \
-o custom-columns=NAME:.metadata.name,\
CPU_REQUEST:.spec.template.spec.containers[].resources.requests.cpu,\
MEMORY_REQUEST:.spec.template.spec.containers[].resources.requests.memory,\
MEMORY_LIMIT:.spec.template.spec.containers[].resources.limits.memory
```

This will display an output similar to the following:

```
NAME                  CPU_REQUEST   MEMORY_REQUEST   MEMORY_LIMIT
fluentd-gcp-v2.0.15   100m          200Mi            300Mi
```

In order to change those values, a [ScalingPolicy][scalingPolicy] needs to be
defined. Currently, only base values are supported (no automatic scaling). The
ScalingPolicy can be created using kubectl. E.g. to set cpu request to 101m,
memory request to 150Mi and memory limit to 400Mi:

```
$ cat <<EOF | kubectl apply -f -
apiVersion: scalingpolicy.kope.io/v1alpha1
kind: ScalingPolicy
metadata:
  name: fluentd-gcp-scaling-policy
  namespace: kube-system
spec:
  containers:
  - name: fluentd-gcp
    resources:
      requests:
      - resource: cpu
        base: 101m
      - resource: memory
        base: 150Mi
      limits:
      - resource: memory
        base: 400Mi
EOF
```

To remove the override and go back to GKE-provided defaults, it is enough to
just remove the ScalingPolicy:

```
$ kubectl delete -n kube-system scalingpolicies.scalingpolicy.kope.io/fluentd-gcp-scaling-policy
```

[cpu]: https://kubernetes.io/docs/tasks/configure-pod-container/assign-cpu-resource/
[memory]: https://kubernetes.io/docs/tasks/configure-pod-container/assign-memory-resource/
[scalingPolicy]: https://github.com/justinsb/scaler