docker-archive/infra-prometheus-consul

Prometheus with dynamic configuration by consul and consul-template

THIS IS DEPRECATED

Prometheus now has native consul support which is the better option than using this docker image.

prometheus-consul

This Docker image run prometheus along with consul-template to update the prometheus config. It adds job, if found, for the following services:

• prometheus-node-exporter
• prometheus-container-exporter
• prometheus-haproxy-exporter
• prometheus-cloudwatch-exporter
• prometheus-pushgateway
• prometheus-alertmanager

It requires the services to have a tag which will be added as label to the job.

As the parent image prom/prometheus, this image stores the metrics in a volume at /prometheus. It's advised to use a data volume container to persist the metrics when updating the image:

$ docker run --name prometheus-data -v /prometheus busybox true
$ docker run --volumes-from prometheus-data dckr/prometheus-consul ...

This way you can remove and recreate the prometheus container while keeping the metrics around.

Configuration

You can pass consul-template parameters directly to the image like this:

$ docker run dckr/prometheus-consul -consul consul:8500

Yo pass command line arguments to prometheus, you can set them via the PROMETHEUS_OPTS env variable:

$ docker run -e PROMETHEUS_OPTS="-alertmanager.url=http://alarm:9095" \
             -p 9090:9090 dckr/prometheus-consul -consul consul:8500

To customize the config template itself or the console templates, create a new directory with the following files (all optional):

./prometheus.conf.tmpl: the prometheus config consul template
./prometheus/consoles: prometheus console templates
./prometheus/console_libraries: prometheus console libraries

Now create a Dockerfile with only FROM dckr/prometheus-consul in it. Building this image will replace the provided template with the templates in your directory.

infra-prometheus-consul's Issues

Issue with reloading configuration - restart rather than reload configuration

The restart of the service should use SIGHUP and not restart the service entirely.

sv restart

restart
Send the commands term, cont, and up to the service, and wait up to 7 seconds for the service to restart. Then report the status or timeout. If the script ./check exists in the service directory, sv runs this script to check whether the service is up and available again; it’s considered to be available if ./check exits with 0.

https://github.com/docker-infra/prometheus-consul/blob/master/consul-template/run#L4

See prometheus documentation for handling configuration reloads.

http://prometheus.io/docs/operating/configuration/