monitoring/metrics in a stack
- A pre-existing docker swarm already setup and configured is necessary for orchestration of the stack.
- A domain or subdomain with the A record pointed to the IP address of the gateway node for automatic issuance of a Let's Encrypt SSL certificate.
mias is a simple passive push-based monitoring/metrics bundle in the form of a docker swarm stack. It can be used to collect system resource metrics like bandwidth, cpu use, etc from remote hosts participating in a docker swarm and then push them back to a centralized location. Alerting then can be enabled and configured within the dashboard to act upon those metrics.
Node-exporter and vmagent are deployed to all docker workers as "agents" via a global service. Preprovisioned instances of Grafana and victoria-metrics-prod are deployed on the node of your choosing via labels to provide an overview and general dashboard.
Simply put, a push-based strategy for delivering metrics was a constraint of this project. If you have recently discovered this project and are looking to use a more common pull-based strategy; we suggest searching for an alternative stack rather than trying to adapt this project to fit your needs.
The mias stack leverages single node, non-replicated, containers of the following services from the latest images below:
- monitoring_grafana "grafana-oss" courtesy of Grafana
- node-exporter courtesy of Prometheus
- vmagent courtesy of VictoriaMetrics
- vmgateway "victoria-metrics-prod" courtesy of VictoriaMetrics
- caddy "Caddy" courtesy of the Caddy Docker Maintainers
No additional external ports are opened beyond ports 443 and 80 of the Grafana container hosting the vmgateway service.
Clone or download this repository. Review ./docker-compose.yml and make any changes that may be required for your production environment.
Specify a single docker worker node to act as the centralized location for the Grafana dashboard.
- Obtain the node ID of a worker. From the manager node of the swarm type:
docker node ls
- Add the
monitoringrole=gatewaylabel to that node:
docker node update --label-add=monitoringrole=gateway <nodeid>
Set a unique Grafana dashboard password in the following file:
./secrets/gf_admin_password.txt
Deploy the stack to all docker worker nodes. From the manager node type:
MIAS_DOMAIN="your-specified-domain.com" docker stack deploy -c docker-compose.yml monitoring
Visit port https://your-specified-domain.com use the username admin with the previously specified password.
You may desire to tag the images within docker-compose.yml instead of relying upon the latest images for a more consistent deployment experience in production.
To review logs, from the manager node of the swarm, type:
docker service logs monitoring_grafana -f
docker service logs monitoring_node-exporter -f
docker service logs monitoring_vmagent -f
docker service logs monitoring_vmgateway -f
docker service logs monitoring_caddy -f
The included Grafana dashboard was modified from the popular Node Exporter Server Metrics by Knut Ytterhaug