Grafana

To display information from our systems we use Grafana to construct dashboards with user defined queries.

More information about our use of Grafana can be found in our infrastructure documentation.

Data sources

To be able to create dashboards, Grafana needs to have access to different datasources.

  • Elasticsearch: Keeps logs from the system and also the audit trail of the users allowing us to generate metrics on request time, request types, number of 4XX requests, user activity etc.

  • Prometheus: Prometheus has a node exporter installed on all the nodes in the Kubernetes cluster. These exporters ships metrics from these nodes. These are used for showing RAM, Disk and CPU usage of individual pods alongside the Kubernetes plugin.

  • Kubernetes Plugin: Gathers information of the Kubernetes cluster available through kubectl alongside Prometheus.

  • Postgres: Grants access to our database and gives us the possibility to query it for things like: number of applications, number of users etc.

  • UCloud metrics (via Prometheus): UCloud exports various metrics via a Prometheus format. These are scraped and made available in Grafana.

Dashboards

We have various dashboards, some examples include:

  • Requests: Here we have 2 different dashboards. One for a general overview of the requests being sent on the cloud system (request time/averages, status codes, number of requests per user) and another to have a more detailed view on a specific request type.

  • Node statistics: A single dashboard that shows the RAM, CPU, Swap and disk usage of all nodes.

  • Postgres Stats: A single dashboard showing stats from the PostgreSQL database (Queries Per Sec., Row stats, number of connections etc.)

  • UCloud dashboard: A single dashboard that shows key numbers of the cloud platform: Number of request, user activity, data stored, jobs information, etc.

  • KPI dashboard: A single dashboard showing Key Performance indicators such as new subscribed users and total amount of users and their respective organization distribution. Jobs activity and which applications are most used.