@SuperQ What are your thoughts on this?

The query id is the primary reason that this collector exists, so, not really useful without it.

@SuperQ could you share a bit of explanation for why you want the query id in metrics (as opposed to logs)?

(What would the benefit be, of exporting a separate time series for each query id - given that the query id generally will never reoccur and thus the values in the time series will never fluctuate?)

This kind of flies directly in the face of Prometheus design and best practice. See, for example, the admonition from the prometheus documentation:

At minimum, do you think it would be prudent to update README-RDS.md to warn the user about the consequences of using the query.yaml in the manner currently advised (particularly since there are also significant financial risks noted in the #549 issue discussion)?

Because in many applications, the number of query ID patterns is fairly limited, and the number of database instance is also much more constrained compared to the number of application servers. This makes the total cardinality acceptable for a lot of use cases.

Knowing the about the performance characteristics of the top X queries can be extremely useful, which is why it's in the example file.

We're planning to remove support for the queries.yaml by moving the various queries into Go code. The whole feature is an anti-pattern that was added to this exporter before it was adopted by the community. As you noticed, people blindly copy the file into production.

I don't think this has anything to do with RDS or not. Any application can produce a lot of series.

What would be OK with me would be to simply comment out the whole query in the yaml file. This way it's not on by default for users that are prone to copy-pasta and don't read through the file to see what it does. I also made some suggested changes to the query but never bothered to update it. That's one of the reasons why we want to eliminate this example file. Good improvements get made locally and never make it upstream.