I wrote [a post a while back](https://nielssegers.be/blog/reverse-proxy-for-local-dns-resolving) about [nginx-proxy](https://github.com/nginx-proxy/nginx-proxy) and how I was using it to resolve local DNS records. Today I decided I wanted to be able to monitor my server remotely using [glances](https://nicolargo.github.io/glances/) running over [Supervisor](http://supervisord.org/) and utilizing glances' internal webserver. 

I could be running `glances` inside a docker container and exposing the docker socket to monitor most of the metrics but that would mean I'd lose useful disk and network information. So since this isn't going to be running inside a container I needed a somewhat creative way to add it to the `local` network.

Install [glances](https://nicolargo.github.io/glances/) and configure [Supervisor](http://supervisord.org/):

```bash
$ sudo apt-get install -y supervisor
$ sudo vi /etc/supervisor/conf.d/glances.conf
```

Configure the program you wish to run through Supervisor:

```
# /etc/supervisor/conf.d/glances.conf
[program:glances]
command=/usr/bin/glances -w
autostart=true
autorestart=true
```

Load and start the configured program:

```bash
$ sudo supervisorctl reread
$ sudo supervisorctl update
```

Confirm everything is running:

```bash
$ sudo supervisorctl status
glances                          RUNNING   pid 647763, uptime 0:35:01
```

Now the creative part, adding it to the `nginx-proxy` network. To be able to do this we are going to make use of a docker image called [docker-host](https://github.com/qoomon/docker-host). Simply explained this image allows you to forward TCP and UDP traffic from within the container to the docker host.

```yaml
version: "3"

services:
  proxy:
    image: jwilder/nginx-proxy:alpine
    environment:
      DEFAULT_HOST: status.local
    ports:
      - 80:80
    volumes:
      - /var/run/docker.sock:/tmp/docker.sock
    restart: always

  glances:
    image: qoomon/docker-host
    expose:
      - 61208
    environment:
      VIRTUAL_HOST: status.local
    restart: always
    cap_add: [ 'NET_ADMIN', 'NET_RAW' ]
```

By giving the `glances` service access to the docker host by giving it the desired Linux capabilities we can expose the `61208` port which is being used by the `glances` webserver and forward all traffic coming in from `status.local` to the correct destination.

Now whenever I visit http://status.local I'm able to monitor my server using the familiar `glances` interface remotely without having to `ssh` each time.

Niels Segers

Full Stack Developer.

Combining Supervisor and nginx-proxy (docker) to remotely monitor Glances