topic Re: Prometheus Extension 2.0 best practices in Extensions

Prometheus Extension 2.0 best practices

SjoerdB — Fri, 25 Oct 2024 12:32:25 GMT

Hi fellow Dynatracers,

I am currently writing an extension for scraping a prometheus endpoint of an appliance. This is a virtual appliance that is under update control of the supplier.

Coding

What is best practice for the extension name (e.g. biz.ictcore.<suppliername>.<supplierProductname>) to use safe namespaces (ictcore.biz is my domain)?
What is best practice for the metrics naming (<suppliername>.<supplierProductname>.<original prometheusMetricname>), to avoid clashes?
This specific Prometheus reply also contains some static values, like a build_info metric. How could I put this information into attributes for my topology, so that it gets visible in the properties (like I already put in some other info in properties)?
```
# HELP node_exporter_build_info A metric with a constant '1' value labeled by version, revision, branch, goversion from which node_exporter was built, and the goos and goarch for the build.
```

example for <suppliername>.<supplierProductname> could be paloalto.cortexxdr or f5.bigip ... etc.

Re: Prometheus Extension 2.0 best practices

SjoerdB — Wed, 04 Dec 2024 13:03:23 GMT

Anyone has an idea on how to get e.g. those attributes from a "static" metric? The other stuff I can just decide myself, for those questions I was only hoping for additional insights from fellow-dynatracers.

Re: Prometheus Extension 2.0 best practices

Mike_L — Wed, 18 Dec 2024 08:57:23 GMT

For question 1 and 2, I think you propose a good system for them.

As for 3, you can always download the YAML's of existing extensions from the Dynatrace hub and have a look. the f5 contain most functionality. Here's how it's done as soon as the value is added as a dimension to the metric line:

topology: types: # Generic Network Types - name: network:device enabled: true displayName: Network device rules: - idPattern: network_device_{device.address} instanceNamePattern: "{instance.name}" iconPattern: f5 sources: - sourceType: Metrics condition: $eq(com.dynatrace.extension.f5.bigip.sys.uptime) requiredDimensions: - key: instance.systemname attributes: - key: devMonitoringMode displayName: Monitoring mode pattern: "{monitoring.mode}"

Re: Prometheus Extension 2.0 best practices

SjoerdB — Mon, 03 Feb 2025 13:16:05 GMT

Took me little longer to put some focus on this, but it "works like a charm", thank you very much for this....!