Clarify what language agents should use for identity#310
Clarify what language agents should use for identity#310breedx-splk wants to merge 2 commits intoopen-telemetry:mainfrom
Conversation
|
Raised this topic at the SIG, the recommendation from the maintainers was to keep the spec generic and create a supplemental document for agent (eg. language instrumentation) specific aspects. @tigrannajaryan @andykellr we'd probably want to move the existing standalone section in context to the Collector out as well in that case? |
|
|
||
| ###### Language Instrumentation Agents | ||
|
|
||
| Instrumentation agents use an OpenTelemetry SDK instance which includes a |
There was a problem hiding this comment.
why not also filter service.name, service.namespace, service.version, service.instance.id to identifying_attributes and everything else to nonidentifying_attributes ?
|
|
||
| For standalone running Agents (such as OpenTelemetry Collector) the following | ||
| attributes SHOULD be specified: | ||
| ###### Standalone Agents |
There was a problem hiding this comment.
@breedx-splk let's move this section and the new "Language Instrumentation Agents" section to a separate document. We can call it something like opentelemetry-guidelines.md and it will contain all considerations that are specific to Otel agents. The specification.md will contain generic protocol specification, which is 99% of the content right now (the only exception is this paragraph, so let's move it out).
|
💨 |
This follows #310. It creates a new document instead of polluting the general high level OpAMP spec with tedious details about actual behavior.
4 participants
The original opamp spec was written with the Collector at top of mind.
There is increasing interest in using OpAMP to manage or re/configure certain aspects of language agents at runtime. This PR adds some clarification about how language instrumentation agents should handle identity and non-identity attributes in OpAMP...effectively just passing along the Resource and nothing more.