Grafana · Your metrics in your own dashboards
Nimbox SRE exposes your metrics as a Prometheus-compatible data source. Connect your own Grafana, write whatever queries you like and build the dashboards you need — no request to us required.
You will only ever see your own data. Every query is scoped to your organisation before it runs, whatever you write.
The quick way: the panels we host
In the console, open Agente and click Abrir mis paneles. It opens with your existing session — no second password — and you get three dashboards already built on your data:
- Infrastructure: hosts up, down, silent and under maintenance; CPU, memory and disk; and the state of your agents.
- Services and incidents: what is down now, how many services were down at once, incidents by severity.
- Service and commitments: whether what was promised is being met — first response and resolution within target.
It is included and there is nothing to install. You can edit them and build your own.
The rest of this page is for those who prefer to connect their own Grafana.
1. Get the credential
In the console, open Agente and scroll to Your metrics in your own Grafana. Click Emitir credencial and give it a name that tells you what it is for — for example, "Operations Grafana".
Shown once
The credential is shown only once, when you create it. We do not store it in a recoverable form: if you lose it, issue another one and revoke the old one. It starts with mx_.
You can hold several at a time, one per tool, and revoke any of them independently. A revoked credential stops working immediately.
2. Add the data source in Grafana
In Grafana go to Connections → Data sources → Add new data source and pick Prometheus.
| Field | Value |
|---|---|
| Prometheus server URL | https://app.nimboxsre.com/api/prom |
| Authentication | (none: a header is used instead) |
| Custom HTTP Headers | Authorization = Bearer YOUR_CREDENTIAL |
Click Save & test. If everything is right, Grafana confirms the source responds.
If it fails
A 401 means the credential is not being sent or is not valid — check the header says Bearer followed by a space and the value. A network error is usually a forgotten https:// in the URL.
3. Query
These are ordinary Prometheus metrics. They support rate(), sum by (host), avg_over_time() and everything else.
| What you want to see | Query |
|---|---|
| Hosts up | monit_host_up |
| CPU per host | monit_cpu_percent |
| Memory used, percent | monit_mem_percent |
| Disk per mount point | monit_disk_percent |
| Services down | monit_service_up == 0 |
| Active incidents by severity | monit_incident_count |
| Incidents resolved by the agent | monit_sre_resolved_total |
| First response within target | monit_sre_response_sla_percent |
Some examples that go a little further:
# The five hosts using the most CPU right now
topk(5, monit_cpu_percent)
# Average memory over the last hour, per host
avg_over_time(monit_mem_percent[1h])
# Disks above 85%
monit_disk_percent > 85
# How many hosts are down
count(monit_host_up == 0)
No need to filter by organisation
Do not add tenant="..." to your queries: it is already done for you. If you name another organisation, the query is rejected with a conflict error rather than quietly returning nothing, so it is clear what happened.
What the credential can do
Read only, and only your series. It cannot write metrics, delete anything, or read your account settings. If it leaks, whoever holds it can see your charts: revoke it from the same screen and issue a new one.
Retention
How much history is available depends on your contract. The included tier keeps 30 days; there are 12-month and 24-month tiers.
From activation onwards
When you extend retention, the archive starts accumulating from that moment. History already outside the previous window is not recovered retroactively.