@jasnell @danlapid this PR attempts to address STOR-4395, but I don't understand the underlying issue well enough to know if this is the right approach in each case – see the TODO comments. Input on that will be useful, I'm also not sure if we'll need to do the same thing for Automated Tracing too – as is, the makeAsyncTraceScope() call only affects the internal tracing system.

I actually don't like context.makeAsyncTraceScope.
Why do we need it now that we have context.attachSpans?

I actually don't like context.makeAsyncTraceScope. Why do we need it now that we have context.attachSpans?

As I understand, this is supposed to be used at the start of all worker events and only pertains to the internal tracing system (so far) – see the internal issue. It is not tied to any specific span. I think that the problem it's trying to address is distinct from the span lifetimes addressed by attachSpans?
I also opened this in part to discuss if we'd need this for user tracing – as I understand it can cause spans to be attributed to the wrong parent span?

Edit: If the span lifetimes are all correct and they end while the IoContext is still available, then it is impossible for them to be attributed to a different worker event.

Updated the PR to only fix this for Hibernate events, adding TODO(later) for things that were unclear to me. This only partially addresses the issue now, but should be wholly uncontroversial at least.

@jclee since you opened the issue for this – this PR is a partial fix for STOR-4395, adding TODOs for the cases that are non-trivial to support. Unfortunately I haven't been able to secure approval for this, lmk if this half-measure works for you if I should close it.