KKP Enterprise Edition (EE) offers optional measuring tools to achieve easier accountability of resources by providing weekly reports about per-cluster CPU and memory utilization. The tool will continuously collect information about all user clusters and create reports containing individual usage values. The configuration and report files can be easily accessed from the dashboard.
The metering tool will be deployed to each seed cluster by the operator.
From there it has access to the user clusters and requests their performance values every five minutes.
The collected information will be written to a CSV file, saved to a
PersistentVolume and uploaded to your S3 bucket.
At the end of the weekly collection period, a CronJob will be triggered so that all data can be evaluated and written to a report file.
All files in the volume are mirrored to a S3 bucket, from where the reports can be accessed.
The dashboard provides a convenient way to list and download all available reports.
Using the dashboard, configuring the Metering tool becomes a breeze. Open the Admin Panel and choose the Metering tab on the left side.
First you need to configure the credentials for your S3 bucket. To do so click on Edit credentials, fill in the credential fields and confirm with the button below.
The next step is to enable metering. Click on Configure Metering, switch on Enable Metering and change the configuration options according to your wishes.
50Gi. See the Kubernetes Docs for a more thorough explanation of valid values
Once the configuration values have been set and the metering has been enabled, the operator will take care of deploying the tool.
Reports will be provided as CSV files.
The file names include the reporting interval including the start and end timestamps, e.g.
While the reports will be stored in your S3-bucket, they can also be accessed from the dashboard. The metering overview provides a list of all reports. Click on the download button on the right side to save a specific report file.
The following values will be written to the reports:
Please note that average values will be calculated over the whole reporting period (one week) and not over the cluster lifespan.
If one of your clusters ran one day and used a single CPU during that time, it will result in a value of
1/7 (approx 0.14) for the field
Average used CPU.
Raw data will be continuously written to CSV files with the name pattern
The timestamp will be the creation date and the metering tool will start a new file at the end of your weekly collection period.
Do not delete these files during a collection period as they are providing the data for the final reports.
The checksum of each record is calculated over the checksum of the previous record and the current record.