I understand that pull-based metric collection ensures ordering since Prometheus centrally controls the scraping time. However, when using OTLP push (e.g., with the OpenTelemetry Collector), data ordering is not guaranteed, and I learned that out_of_order_time_window can be configured to handle this issue.

What I’m still unclear about is how this configuration helps resolve out-of-order ingestion. I’d truly appreciate it if someone could help me understand the mechanism behind it, as I’m still learning.

Thank you in advance!