14768 pending time executors approach 2

[mateusz-nalepa]

Draft for #14768

Metric name: executor.queue.wait.duration
Type: Histogram
Unit: s
Brief: Measures how long a task waits in an executor queue before being executed by a thread

Attributes:

• thread.name (string, optional): Name of thread that executes the task

Notes:

• This metric helps identify saturation in thread pools
• Measurement starts when the task is enqueued to the executor
• Tasks scheduled with delay (e.g. via ScheduledThreadPoolExecutor) do not trigger metric until they are actually enqueued

[trask]

@mateusz-nalepa can you summarize in the description above the new telemetry that is emitted by this instrumentation?

[mateusz-nalepa]

@trask done
Please let me know if this description is enough

[trask]

@mateusz-nalepa more I was looking for a description of the metrics from semantic convention perspective

metric name, unit, brief, notes, attributes, etc

[mateusz-nalepa]

@trask
I've updated description
I was trying my best based on https://opentelemetry.io/docs/specs/semconv/general/metrics/ and https://opentelemetry.io/docs/specs/semconv/http/http-metrics/#metric-httpserverrequestduration
Please let me know if updated description is enough

[trask]

great, thanks! that helps me a lot to be able to review it just from semantic convention perspective

Unit: ms

You'll want to change this to s (see https://github.com/open-telemetry/semantic-conventions/blob/main/docs/general/metrics.md#instrument-units)

Attributes:

• thread.name (string): Name of thread that executes the task

this will need to be opt-in if you want to capture it, since it could be high cardinality

[mateusz-nalepa]

@trask fixed
From my perspective, I wasn't expecting this PR to have any review right now, when it's in draft status, so I know, that it will be an amazing journey to work on it!