It seems the container-exporter used in a docker-container (Docker 1.9.1) is using much more mem than its cousins.

Query: avg(container_memory_usage_bytes{name=~"prom.*"}  / 1024 / 1024) by (name)
Element                                 Value
{name="prometheus-consul-exporter"}     10.352864583333334
{name="prometheus-haproxy-exporter"}    11.6328125
{name="prometheus-container-exporter"}  325.6673828125
{name="prometheus-node-exporter"}       13.4107421875
{name="prometheus-statsd-exporter"}     10.933203125
{name="prometheus"}                     5294.07421875

CONTAINER                       CPU %               MEM USAGE / LIMIT     MEM %               NET I/O               BLOCK I/O
prometheus-container-exporter   0.00%               343.7 MB / 8.372 GB   4.11%               12.43 MB / 123.6 MB   4.1 MB / 0 B
prometheus-node-exporter        0.00%               15.16 MB / 8.372 GB   0.18%               0 B / 0 B             4.305 MB / 0 B
prometheus-statsd-exporter      0.18%               11.92 MB / 8.372 GB   0.14%               1.243 GB / 68.41 MB   5.325 MB / 0 B

Uptime here is 10 days for all prometheus-containers, and scrape time of 60 sec.

I will try using cAdvisor instead of the container-exporter, but logging here in case someone takes up the maintenance of this repo.

Hello,

I'm using container-exporter to monitor docker containers on debian 8 which use systemd as init system
Also I see system services such as sshd or mdadm.
Example:
`
container_memory_working_set_bytes{id="/system.slice/docker-3f7eba3c3b2a63c8bbba4f679ca2d2ba704b501890c112218dbdfaa66e2a8041.scope",name="admiring_newton"} 0

container_memory_working_set_bytes{id="/system.slice/docker-576a59ee026bf956a550cf1cd2beb42c1859a3ae44b110b5fbc504cf2127395d.scope",name="cadvisor"} 0
container_memory_working_set_bytes{id="/system.slice/docker-6fcea541f42e9de0778f5e543b145a0e9a845048fabc66829e045ddf276820b5.scope",name="graylog-web-ui"} 0

container_memory_working_set_bytes{id="/system.slice/docker-7b66c376e6db9bfd000fe750a9dd7c6f97b2a86b3eac852fb8aab3a970a4f37f.scope",name="sinopia"} 0
`

export docker labels

Now that Docker 1.6 has labels, it would be awesome if this exporter could pick those up. Otherwise, I have no way of differentiating different types of container services as I don't name them (I use Consul SERVICE_NAME env var)

Hi all:

I have some questions,hope anyone can help me.

I notice that monitor container now use cAdvisor, but i don't konw why cadvisor instead of container_exporter?

If i don't use cAdvisor,is there another way to monitor docker container?

Thanks.

Hello,

I have a problem on centos 7 : all stats are null.
I have try launch with "-v /sys/fs/cgroup:/cgroup" with no parent. Here is two tests :

./container-exporter -parent='/inexistent_path'
2015/02/02 10:22:44 Starting Server: :8080

No problem... all containers are displayed on :8080... but no values... (all are "0")

And now without a "/" :

./container-exporter -parent='inexistent_path'
2015/02/02 10:26:40 Starting Server: :8080
2015/02/02 10:26:42 Error reading container stats: open /cgroup/cpuacct/system.slice/docker-7543eec9fafc0ba3bd1892b684ddab2af0e760debb893b61d9da1e3736e8f2af.scope/inexistent_path/7543eec9fafc0ba3bd1892b684ddab2af0e760debb893b61d9da1e3736e8f2af/cpuacct.stat: no such file or directory

Why the container.ID is append again to "/cgroup/cpuacct/system.slice/docker-7543eec9fafc0ba3bd1892b684ddab2af0e760debb893b61d9da1e3736e8f2af.scope/{$parent}/ ?

Thx in advance for any clue/answer :)

no stats are collected on coreos
started as
{code}
docker run -d --name container_exporter -p 9104:9104 -v /sys/fs/cgroup:/cgroup -v /var/run/docker.sock:/var/run/docker.sock prom/container-exporter:latest
{code}

example
{code}

TYPE container_cpu_usage_seconds_total counter

container_cpu_usage_seconds_total{id="76ad37b466443f963486695e8c7270c56d62d2238513567b3e6cf927311e36d8",image="prom/container-exporter:latest",name="container_exporter",type="kernel"} 0
container_cpu_usage_seconds_total{id="76ad37b466443f963486695e8c7270c56d62d2238513567b3e6cf927311e36d8",image="prom/container-exporter:latest",name="container_exporter",type="user"} 0
{code}

I suspect this is having issue with the cgroups are rearranged on coreos
example of cgroup on most systems
{code}
/sys/fs/cgroup/docker/%(id)/cgroup.procs
{code}
in coreos
{code}
/sys/fs/cgroup/systemd/system.slice/docker-%(id).scope/cgroup.procs
{code}

I can only see cpu and memory metrics. Don't see expected values as per git codebase. Maybe I missed a config or other tweak. Can you please help me understand ?

It would be cool if there were a thinner container-exporter image. The go:onbuild base is currently 514MB and the layer that does the go-wrapper download is 72MB. The layer that does the go install itself is only 16MB, so I suspect a resulting thin image could be around that size.

The swarm image is a good example of this pattern: https://github.com/docker/swarm-library-image/blob/master/Dockerfile

I've also created a couple of them myself: https://github.com/synctree/docker-logstash-forwarder https://github.com/md5/docker-gen/tree/static-build-image

I have observed that the container_exporter crashed because of to many open filehandles.
When you have a look at the metric "process_open_fds" you can see that it increases over time.

Can you help to find out what is wrong?

I have tested container_exporter within docker and outside of docker with the same result.

Not sure how/where to make this request/suggestion, but network usage metrics may prove useful when collecting metrics on containers.