Skip to main content

cAdvisor API


cAdvisor (Container Advisor) provides container users an understanding of the resource usage and performance characteristics of their running containers.

It is a running daemon that collects, aggregates, processes, and exports information about running containers.

This Pack aims to monitor metrics exposed through cAdvisor API endpoint.

Monitored objects​

This Pack contains one Host Template and several Service Templates to monitor node resource allocation and container performances:

  • Container-Usage
  • Container-Disk-IO
  • Container-Traffic
  • Node-Status

Discovery rules​

The cAdvisor API Pack comes with several service discovery rules.

To monitor container metrics, you must launch a discovery to add relevant services:

Cloud-cAdvisor-API-Container-Disk-IODiscover containers and monitor Disk-IO metrics
Cloud-cAdvisor-API-Container-UsageDiscover containers and monitor CPU & RAM consumption
Cloud-cAdvisor-API-Container-TrafficDiscover containers and monitor bandwidth utilization

Collected metrics​

Metric nameDescriptionUnit
<container_name_or_id>#container.cpu.countNumber of allocated CPU(s)count
<container_name_or_id>#container.cpu.utilization.percentageCPU utilization%
<container_name_or_id>#container.cpu.user.utilization.percentageCPU User utilization%
<container_name_or_id>#container.cpu.system.utilization.percentageCPU System utilization%
<container_name_or_id>#container.memory.usage.bytesMemory usageB
<container_name_or_id>#container.memory.cache.usage.bytesCached memory usageB
<container_name_or_id>#container.memory.rss.usage.bytesRSS memory usageB
<container_name_or_id>#container.swap.usage.bytesSWAP memory usageB

By default, the --use-name flag will instance the metric with the <container_name>. If you want to use the <container_id> instead, remove it from the EXTRAOPTIONS macro at the Service Template level.



A running cAdvisor container should be available. You can refer to the official quick start.

Network flow​

The Poller should be able to reach the cAdvisor Host over the TCP/8080 port. Note that the port may be different on your setup.


  1. Install the Centreon Plugin package on every Centreon Poller expected to monitor containers with cAdvisor:
yum install centreon-plugin-Cloud-cAdvisor-Api
  1. On the Centreon Web interface, install the cAdvisor API Plugin Pack through the Configuration > Plugin Packs > Manager page.

Host configuration​

Add a host from the Configuration > Hosts menu and select the Cloud-cAdvisor-API template.

Here is a quick description of available configuration Macros:

MandatoryMacroDescriptionDefault value
xCADVISORAPIPROTOProtocol used to talk with the cAdvisor APIhttp
xCADVISORAPIPORTNetwork port cAdvisor API listens over8080
xCADVISORAPIPATHAPI Path to container metrics information/containers/docker/
CADVISORAPIEXTRAOPTIONSExtraoptions you may want to add to your command--http-backend=curl --insecure
PROXYURLURL of a proxy to use to reach cAdvisor API

Click on the Save button and you're good to push the configuration to the Engines.

How to test my plugin and what do the main parameters stand for?​

Once the plugin is installed, you can test it by logging into the CLI with the centreon-engine user.

/usr/lib/centreon/plugins// \
--plugin=cloud::cadvisor::restapi::plugin --mode=traffic \
--hostname= --port='8080' --proto='http' \
--api-path='/containers/docker/' --http-backend=curl --insecure \
--container-id='' --container-name='' --filter-name='^gray$' \
--warning-traffic-in='' --critical-traffic-in='' \
--warning-traffic-out='' --critical-traffic-out='' \
--verbose --use-name

Expected output is shown below:

OK: Container 'gray.eth0' Traffic In: 43.99 b/s, Traffic Out: 39.92 b/s | ''=43.99b/s;;;0; 'gray.eth0#container.traffic.out.bitspersecond'=39.92b/s;;;0;
Container 'gray.eth0' Traffic In: 43.99 b/s, Traffic Out: 39.92 b/s

The command above checks the incoming and outgoing traffic for a container (--plugin=cloud::cadvisor::restapi::plugin --mode=traffic). The focus is on the gray (--filter-name='^gray$') container and we use this name as the perfdata instance --use-name.

All available options for a given mode can be displayed by adding the --help parameter to the command:

/usr/lib/centreon/plugins// \
--plugin=cloud::cadvisor::restapi::plugin --mode=traffic \


UNKNOWN: curl perform error : Timeout was reached​

When this error occurs, check that you can reach the cAdvisor over the 8080/HTTP port.

You can use the --debug flag to obtain additional information about the reason of the failure.

If a proxy is required, add its URL within the PROXYURL Macro at the Host level.

UNKNOWN: No containers found or no data available for this specific metric.​

This error means that no data is available for this specific metric or that the container name used in the filter might be mispelled (this can't happen if you use the service discoveru feature).