Collectd within a Docker image
Clone or download

README.md

docker-collectd

Description

collectd is a daemon which collects system performance statistics periodically and provides mechanisms to store the values in a variety of ways, for example in RRD files.

This image allows you to run collectd in a completely containerized environment, but while retaining the ability to report statistics about the host the collectd container is running on.

The SignalFx collectd Docker plugin is also included, and will report metrics about your containers.

Image Options

Base Image Image Name
Ubuntu quay.io/signalfuse/collectd
Alpine Linux quay.io/signalfuse/collectd-alpine

How to use this image

Run collectd with the default configuration with the following command:

docker run --privileged \
  --net="host" \
  -e "SF_API_TOKEN=XXXXXXXXXXXXXXXXXXXXXX" \
  -v /:/hostfs:ro \
  -v /var/run/docker.sock:/var/run/docker.sock \
  quay.io/signalfuse/collectd

If you don't want to report Docker metrics from the SignalFx Docker collectd plugin, then you can leave out the argument that mounts the Docker socket -v /var/run/docker.sock:/var/run/docker.sock.

docker run --privileged \
  --net="host" \
  -e "SF_API_TOKEN=XXXXXXXXXXXXXXXXXXXXXX" \
  -v /:/hostfs:ro \
  quay.io/signalfuse/collectd

If you just want to look around inside the image, you can run bash. Then you can run /opt/setup/run.sh yourself to configure the bind mount and start collectd.

docker run -ti --privileged \
  --net="host" \
  -e "SF_API_TOKEN=XXXXXXXXXXXXXXXXXXXXXX" \
  -v /:/hostfs:ro \
  -v /var/run/docker.sock:/var/run/docker.sock \
  quay.io/signalfuse/collectd bash

If you don't want to pass your API token through a command-line argument, you can put SF_API_TOKEN=XXXXXXXXXXXXXXXXXXXXXX into a file (that you can chmod 600) and use the --env-file command-line argument:

docker run --privileged --env-file=token.env \
  --net="host" \
  -v /var/run/docker.sock:/var/run/docker.sock \
  -v /:/hostfs:ro \
  quay.io/signalfuse/collectd

If you want to apply custom dimensions and values to all metrics, you set the environment variable DIMENSIONS to a string of space separated Key/Value pairs defined KEY=VALUE.

docker run --privileged --env-file=token.env \
  --net="host" \
  -e DIMENSIONS="hello=world foo=bar"
  -v /var/run/docker.sock:/var/run/docker.sock \
  -v /:/hostfs:ro \
  quay.io/signalfuse/collectd

On CoreOS because /etc/*-release are symlinks you want to mount /usr/share/coreos in place of /etc.

docker run --privileged \
  --net="host" \
  -e "SF_API_TOKEN=XXXXXXXXXXXXXXXXXXXXXX" \
  -v /:/hostfs:ro \
  -v /var/run/docker.sock:/var/run/docker.sock \
  quay.io/signalfuse/collectd

If you're using the dogstatsd listener in the SignalFx plugin. You will need to expose the container port and specify the port with an environment variable.

docker run --privileged \
  --net="host" \
  -e "SF_API_TOKEN=XXXXXXXXXXXXXXXXXXXXXX" \
  -e DOG_STATSD_PORT=8126 \
  -p 8126:8126 \
  -v /var/run/docker.sock:/var/run/docker.sock \
  -v /:/hostfs:ro \
  quay.io/signalfuse/collectd

FAQ

Do I need to run the container as privileged?

Yes. Collectd needs access to the parent host's /proc filesystem to get statistics. It's possible to run collectd without passing the parent host's /proc filesystem without running the container as privileged, but the metrics would not be accurate.

Do I need to provide the network flag?

Yes. Setting the --net option to "host" allows the container to see the host's networking stack and provide accurate information.

Do I need to mount the host's / to /hostfs?

This is a requirement for the collectd to accurately report information about the underlying host.

Do I need to provide a host name?

You may set FQDN_LOOKUP to true to enable collectd to perform an FQDN look up, but this may not work in all circumstances. If a lookup fails the container will exit and log a message indicating that fqdn lookup failed. You may optionally set the environment variable using the runtime option -e "COLLECTD_HOSTNAME=<hostname>" where <hostname> is a valid host name string. You may also set the environment variable in an optional environment file. If a host name is not specified /hostfs/etc/hostname name will be used. The container will exit with an error if FQDN_LOOKUP is not true, COLLECTD_HOSTNAME is not set, and hostname can't be found under /hostfs/etc/hostname.

Can I configure anything?

Yes! You are required to set the SF_API_TOKEN, but you also can set the following:

Environment Variable Description Example
COLLECTD_BUFFERSIZE If set we will set write_http's buffersize to the value provided, otherwise a default value of 16384 will be used. -e "COLLECTD_BUFFERSIZE=<size>"
COLLECTD_CONFIGS If set we will include $COLLECTD_CONFIGS/*.conf in collectd.conf where you can include any other plugins you want to enable. These of course would need to be mounted in the container with -v. -e "COLLECTD_CONFIGS=<path to confs>"
COLLECTD_FLUSHINTERVAL if set we will set write_http's flush interval to the value provided, otherwise a default value of what COLLECTD_INTERVAL is set to will be used. -e "COLLECTD_FLUSHINTERVAL=<interval>"
FQDN_LOOKUP=true If set to [ True, true, or TRUE ], then collectd will be configured to use FQDNLook up to identify the value for host.
COLLECTD_HOSTNAME If set and FQDN_LOOKUP is unset or false, then we will set the collectd hostname to the supplied value in /etc/collectd/collectd.conf. If the environment variable is not set, we will attempt to cat /mnt/hostname for the host's hostname. If no hostname is discovered, we will exit with an error code of 1 and display a message indicating that a hostname could not be found. -e "COLLECTD_HOSTNAME=<hostname>"
USE_AWS_UNIQUE_ID_AS_HOSTNAME If non-empty, the hostname will be set to the AWS Unique ID, which is determined using the EC2 metadata endpoint. This is useful in ECS clusters where the container's hostname might be localhost. Note that you will also have a dimension called AWSUniqueId with the same value as the host dimension. This is necessary duplication because we do property syncing to the AWSUniqueId dimension and not to host. -e USE_AWS_UNIQUE_ID_AS_HOSTNAME=true
COLLECTD_INTERVAL If set we will use the specified interval for collectd and the plugin, otherwise the default interval is 10 seconds. -e "COLLECTD_INTERVAL=<interval>"
WRITE_HTTP_TIMEOUT If set we will set write_http's timeout to the value provided, otherwise the default value of WRITE_HTTP_TIMEOUT will be 9000. -e WRITE_HTTP_TIMEOUT=9000
LOG_HTTP_ERROR If set we will set write_http's LogHttpError settting to the value provided, otherwise the default value of LOG_HTTP_ERROR will be false. -e LOG_HTTP_ERROR=false
DIMENSIONS If set with a string of space separated key=value dimension pairs, then each metric dipatched will be tagged with those dimensions. -e DIMENSIONS="foo=bar hello=world"
DISABLE_AGGREGATION If set to [ True, true, or TRUE ], disables the collectd aggregation plugin -e "DISABLE_AGGREGATION=True"
DISABLE_CPU If set to [ True, true, or TRUE ], disables the collectd cpu plugin -e "DISABLE_CPU=True"
DISABLE_CPUFREQ If set to [ True, true, or TRUE ], disables the collectd cpu frequency plugin -e "DISABLE_CPUFREQ=True"
DISABLE_DF If set to [ True, true, or TRUE ], disables the collectd df plugin -e "DISABLE_DF=True"
DISABLE_DISK If set to [ True, true, or TRUE ], disables the collectd disk plugin -e "DISABLE_DISK=True"
DISABLE_DOCKER If set to [ True, true, or TRUE ], disables the collectd docker plugin -e "DISABLE_DOCKER=True"
DISABLE_HOST_MONITORING If set to [ True, true, or TRUE ], disables the collectd aggregation, cpu, cpu frequency, df, disk, docker, interface, load, memory, protocols, vmem, uptime, SignalFx collectd, and write http plugins -e "DISABLE_HOST_MONITORING=True"
DISABLE_INTERFACE If set to [ True, true, or TRUE ], disables the collectd interface plugin -e "DISABLE_INTERFACE=True"
DISABLE_LOAD If set to [ True, true, or TRUE ], disables the collectd load plugin -e "DISABLE_LOAD=True"
DISABLE_MEMORY If set to [ True, true, or TRUE ], disables the collectd memory plugin -e "DISABLE_MEMORY=True"
DISABLE_PROTOCOLS If set to [ True, true, or TRUE ], disables the collectd protocols plugin -e "DISABLE_PROTOCOLS=True"
DISABLE_VMEM If set to [ True, true, or TRUE ], disables the collectd vmem plugin -e "DISABLE_VMEM=True"
DISABLE_UPTIME If set to [ True, true, or TRUE ], disables the collectd uptime plugin -e "DISABLE_UPTIME=True"
DISABLE_SFX_PLUGIN If set to [ True, true, or TRUE ], disables the SignalFx collectd plugin -e "DISABLE_SFX_PLUGIN=True"
DISABLE_WRITE_HTTP If set to [ True, true, or TRUE ], disables the collectd write http plugin -e "DISABLE_WRITE_HTTP=True"
DISABLE_AGENT_PROCESS_STATS If set to [True, true, or TRUE ], disables metrics about the collectd process -e "DISABLE_AGENT_PROCESS_STATS=True"
PER_CORE_CPU_UTIL If set to [True, true, or TRUE], configures the SignalFx collectd plugin to report CPU utilization per core -e "PER_CORE_CPU_UTIL=True
DOG_STATSD_PORT If specified, configures the SignalFx collectd plugin to read DogStatsdD metrics over the configured port. You must also expose this port on the container. -e "STATSD_PORT=8126
DOCKER_TIMEOUT The number of seconds calls to the Docker API should wait to timeout -e "DOCKER_TIMEOUT=3"
DOCKER_INTERVAL If set we will use the specified interval for the docker plugin, otherwise the global collectd interval (defaulted to 10 secs) will be used -e "DOCKER_INTERVAL=10"
SF_INGEST_HOST Will set the ingest url on the signalfx metadata plugin and on collectd itself -e "SF_INGEST_HOST=http://ingest.proxy.com"

Extending

Click this link to read about Extending SignalFx Docker collectd image