| meta | content | tags | dates | categories | |||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
|
|
observability cockpit retention metrics logs |
|
|
This page explains how to fine-tune your metrics ingestion rate while using Cockpit. Find out more about Cockpit pricing in the dedicated documentation.
If you wish to avoid extra costs, you can define a limit to the number of samples of metrics you send to Cockpit. The following table shows an estimate of the pricing (€0.00000015 per sample of metrics) based on the push rate of the sample:
| Sample ingestion rate per seconds | Samples ingested per day | Price per day | Samples ingested per month | Price per month |
|---|---|---|---|---|
| 1 | 86 400 | €0.01296 | 2 678 400 | €0.4017 |
| 10 | 864 000 | €0.1296 | 26 784 000 | €4.0176 |
| 100 | 8 640 000 | €1.296 | 267 840 000 | €40.176 |
| 1000 | 86 400 000 | €12.96 | 2 678 400 000 | €401.76 |
Within the Cockpit dashboard on Grafana, click Cockpit Overview. The panel named Metrics ingestion rate displays the ingestion rate for samples coming from Scaleway products, and the ingestion rate for your own samples if you look at "Other Metrics".
If you are using your own Grafana to visualize metrics, use the following query sum(rate(observability_cockpit_ingestion_samples_total{is_from_scaleway="false"}[5m])) OR on() vector(0) to know what your current ingestion rate is.
A common good practice to control the number of samples you send to Cockpit, is to have a scrape_interval of 1 minute.
Let us say you have 1000 samples exposed and a scrape_interval of 10s. This means you will have a rate of 100 sample/s and pay around €40 per month.
If you increase the scrape_interval to 60s, you will have a rate of 1,66 sample/s, which amounts to around €6.69 per month.
If you are using Prometheus to remote write metrics to your Cockpit, you can tweak the scrape_interval using the following configuration:
```yaml
global:
scrape_interval: 60s
scrape_configs:
...
remote_write:
- url: https://xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx.metrics.cockpit.fr-par.scw.cloud/api/v1/push
headers:
"X-Token": <COCKPIT_TOKEN_SECRET_KEY>
```
Here is an example of a Grafana agent configuration file with a scrape interval of 60 seconds:
```yaml
metrics:
global:
scrape_interval: 60s
remote_write:
- url: https://xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx.metrics.cockpit.fr-par.scw.cloud/api/v1/push
headers:
"X-Token": <COCKPIT_TOKEN_SECRET_KEY>
integrations:
...
```
Another good practice to control the number of samples you send to Cockpit, is to only send metrics that you want into your Cockpit, by filtering what you send.
Here is a configuration example with Prometheus:
```yaml
...
remote_write:
- url: https://xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx.metrics.cockpit.fr-par.scw.cloud/api/v1/push
headers:
"X-Token": <COCKPIT_TOKEN_SECRET_KEY>
write_relabel_configs:
- source_labels: [__name__]
regex: 'my_app_metrics(.*)'
action: keep
```
This configuration will only send metrics starting with my_app_metrics to your Cockpit.
For more information, refer to the Prometheus documentation.
Many agents, such as cadvisor and node_exporter expose a lot of metrics by default.
The following metrics collectors are enabled by default on a node_exporter. You can configure them to filter what is sent to your Cockpit.
Use the following configuration example for a Grafana Alloy agent with the prometheus.exporter.unix integration. For more information on how to configure and use the Grafana Alloy agent, refer to the dedicated documentation.
```yaml
...
prometheus.exporter.unix "node" {:
set_collectors:
# Anything not provided in the list below will be disabled by default
- uname
- cpu
- cpufreq
- loadavg
- meminfo
- filesystem
- netdev
```