# This configuration file for Fluentd / td-agent is used
# to watch changes to Docker log files that live in the
# directory /var/lib/docker/containers/ and are symbolically
# linked to from the /varlog directory using names that capture the
# pod name and container name. These logs are then submitted to
# Google Cloud Logging which assumes the installation of the cloud-logging plug-in.
#
# Example
# =======
# A line in the Docker log file might like like this JSON:
#
# {"log":"2014/09/25 21:15:03 Got request with path wombat\n",
#  "stream":"stderr",
#   "time":"2014-09-25T21:15:03.499185026Z"}
#
# Currently, the log information is ingested as plain text rather than JSON.
# TODO: When Cloud Logging supports it, ingest as JSON.
#
# The record reformer is used to write the tag to focus on the pod name
# and the Kubernetes container name. For example a Docker container's logs
# might be in the directory:
#  /var/lib/docker/containers/997599971ee6366d4a5920d25b79286ad45ff37a74494f262e3bc98d909d0a7b
# and in the file:
#  997599971ee6366d4a5920d25b79286ad45ff37a74494f262e3bc98d909d0a7b-json.log
# where 997599971ee6... is the Docker ID of the running container.
# The Kubernetes kubelet makes a symbolic link to this file on the host machine
# in the /var/log/containers directory which includes the pod name and the Kubernetes
# container name:
#    synthetic-logger-0.25lps-pod_default-synth-lgr-997599971ee6366d4a5920d25b79286ad45ff37a74494f262e3bc98d909d0a7b.log 
#    -> 
#    /var/lib/docker/containers/997599971ee6366d4a5920d25b79286ad45ff37a74494f262e3bc98d909d0a7b/997599971ee6366d4a5920d25b79286ad45ff37a74494f262e3bc98d909d0a7b-json.log
# The /var/log directory on the host is mapped to the /varlog directory in the container
# running this instance of Fluentd and we end up collecting the file:
#   /varlog/containers/synthetic-logger-0.25lps-pod_default-synth-lgr-997599971ee6366d4a5920d25b79286ad45ff37a74494f262e3bc98d909d0a7b.log
# This results in the tag:
#  varlog.containers.synthetic-logger-0.25lps-pod_default-synth-lgr-997599971ee6366d4a5920d25b79286ad45ff37a74494f262e3bc98d909d0a7b.log
# The record reformer is used is discard the varlog.containers prefix and
# the Docker container ID suffix and "kubernetes." is pre-pended giving the
# final tag which is ingested into Elasticsearch:
#   kubernetes.synthetic-logger-0.25lps-pod_default-synth-lgr
# This makes it easier for users to search for logs by pod name or by
# the name of the Kubernetes container regardless of how many times the
# Kubernetes pod has been restarted (resulting in a several Docker container IDs).
# TODO: Propagate the labels associated with a container along with its logs
# so users can query logs using labels as well as or instead of the pod name
# and container name.

<source>
  type tail
  format json
  time_key time
  path /varlog/containers/*.log
  pos_file /varlog/gcp-containers.log.pos
  time_format %Y-%m-%dT%H:%M:%S
  tag reform.*
  read_from_head true
</source>

<match reform.**>
  type record_reformer
  enable_ruby true
  tag kubernetes.${tag_suffix[3].split('-')[0..-2].join('-')}
</match>

<source>
  type tail
  format none
  path /varlog/salt/minion
  pos_file /varlog/gcp-salt.pos
  tag salt
</source>

<source>
  type tail
  format none
  path /varlog/startupscript.log
  pos_file /varlog/gcp-startupscript.log.pos
  tag startupscript
</source>

<source>
  type tail
  format none
  path /varlog/docker.log
  pos_file /varlog/gcp-docker.log.pos
  tag docker
</source>

<source>
  type tail
  format none
  path /varlog/etcd.log
  pos_file /varlog/gcp-etcd.log.pos
  tag etcd
</source>

<source>
  type tail
  format none
  path /varlog/kubelet.log
  pos_file /varlog/gcp-kubelet.log.pos
  tag kubelet
</source>

<source>
  type tail
  format none
  path /varlog/kube-apiserver.log
  pos_file /varlog/gcp-kube-apiserver.log.pos
  tag kube-apiserver
</source>

<source>
  type tail
  format none
  path /varlog/kube-controller-manager.log
  pos_file /varlog/gcp-kube-controller-manager.log.pos
  tag kube-controller-manager
</source>

<source>
  type tail
  format none
  path /varlog/kube-scheduler.log
  pos_file /varlog/gcp-kube-scheduler.log.pos
  tag kube-scheduler
</source>

<match **>
  type google_cloud
  # Set the chunk limit conservatively to avoid exceeding the GCL limit
  # of 2MB per write request.
  buffer_chunk_limit 512K
  # Cap buffer memory usage to 512KB/chunk * 128 chunks = 65 MB
  buffer_queue_limit 128
  flush_interval 5s
  # Never wait longer than 5 minutes between retries.
  max_retry_wait 300
  # Disable the limit on the number of retries (retry forever).
  disable_retry_limit
</match>