-
Notifications
You must be signed in to change notification settings - Fork 469
Description
Customers rely heavily on custom dimensions/tags in request telemetry to enrich logs and metrics with additional context. These dimensions are used for various purposes, including:
- Correlating requests across distributed systems
- Adding tenant or user-specific metadata
- Tracking feature flags or experiment groups
- Debugging and filtering telemetry based on custom logic
In the in-process model, since both the host and customer code run in the same process, customers can directly enrich RequestTelemetry with custom dimensions.
However, in the isolated worker model (and other out-of-process language workers), this capability is lost because the worker and host run in separate processes. As a result, customers cannot directly modify the RequestTelemetry generated by the host.
Proposal
Implement a reverse propagation mechanism that allows the worker process to send custom dimensions/tags to the host process. The host can then attach these dimensions to the RequestTelemetry before it's sent to Application Insights.