Clarity around connector/spanmetrics dimensions_cache_size

[danelson]

Component(s)

connector/spanmetrics

Describe the issue you're reporting

The spanmetrics connector exposes a dimensions_cache_size option. I am curious about the following:

• What is the behavior when dimensions_cache_size is reached?
• Are dimensions ever reclaimed when they don't receive updates?
  • The use case I am trying to solve for is when I see a greate number of dimensions than dimensions_cache_size but over a long time horizon. Similar to Ability to forget and reclaim MetricPoint (Delta) opentelemetry-dotnet#2360

I also think that the README should be updated to address these questions.

[github-actions]

Pinging code owners:

• connector/spanmetrics: @albertteoh @kovrus

See Adding Labels via Comments if you do not have permissions to add labels yourself.

[danelson]

I was under the working assumption that dimensions_cache_size was going to prevent dimensions from being exported but that doesn't appear to be the case. Is the setting only affecting performance and not functionality?

[kovrus]

Is the setting only affecting performance and not functionality?

yes, it is used only to improve performance.

What is the behavior when dimensions_cache_size is reached?

Old items in a cache will be invalidated

[danelson]

@kovrus thank you. I was worried that having very high dimensionality was going to result in me missing metrics. I have what I need from this so I am closing.