Docker daemon at unix://var/run/docker.sock does not support container statistics!

[pavan530]

I m getting this issue in collectd logs after deploying docker-collectd-plugin. do anyone knows the problem?

[dennisxrow]

Hello,
you may have to enable the docker API by passing these options to your docker deamon:

-H tcp://127.0.0.1:4243 -H unix:///var/run/docker.sock

If this wont work for you, you can try to change the BaseURL config to something like this:

<Module dockerplugin>
    BaseURL "http://localhost:4243"
    Timeout 10
</Module>

This fixed the issue for me on Centos7, you may want to check out this link too:
http://www.virtuallyghetto.com/2014/07/quick-tip-how-to-enable-docker-remote-api.html

[korablin]

@dennisxrow
+1
Hi! I am having the same problem, and apparently your solution doesn't work for me.

Do you have any other ideas how it can be fixed?
I am using RHEL 7.1

[dennisxrow]

@korablin We are currently not using this plugin anymore, but you may want to check if you docker api is actually reachable.

In example curl localhost:4243/containers/json on the machine where the collectd client is running, it should return some json.

[babatundebusari]

@dennisxrow
so what plugin are you using to collect docker containers metrics for collectd?

[melhesedek]

This issue is still open so I'll share some observations.

The Docker daemon at unix://var/run/docker.sock does not support container statistics! warning may also be a result of a SELinux policy blocking access to the sock file, a random glitch when API somehow couldn't respond within configured timeout, or a situation when collectd is started before Docker is properly initialized on system startup. In all these cases self.client.version() call throws an exception resulting in this warning.