Detect if listener is already wrapped by once

[dyladan]

Fixes #2971

This is an alternative solution to #3123

[codecov]

Codecov Report

Merging #3122 (316ccae) into main (df58fac) will increase coverage by 0.14%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##             main    #3122      +/-   ##
==========================================
+ Coverage   93.07%   93.21%   +0.14%     
==========================================
  Files         195      186       -9     
  Lines        6384     5838     -546     
  Branches     1347     1252      -95     
==========================================
- Hits         5942     5442     -500     
+ Misses        442      396      -46     

Impacted Files	Coverage Δ
...sync-hooks/src/AbstractAsyncHooksContextManager.ts	97.10% <100.00%> (+0.08%)	⬆️
...y-instrumentation-grpc/src/enums/AttributeNames.ts
...entelemetry-instrumentation-grpc/src/grpc/index.ts
...ry-instrumentation-grpc/src/grpc-js/serverUtils.ts
...metry-instrumentation-grpc/src/grpc/clientUtils.ts
...es/opentelemetry-instrumentation-grpc/src/utils.ts
...elemetry-instrumentation-grpc/src/grpc-js/index.ts
...ry-instrumentation-grpc/src/grpc-js/clientUtils.ts
...emetry-instrumentation-grpc/src/instrumentation.ts
...metry-instrumentation-grpc/src/grpc/serverUtils.ts

[vmarchaud]

I prefer this solution over #3123 since it only change patch for add listener (whereas the other one modify the bind which is much more used).

[Flarna]

be aware that OTel context manager is likely not be the only one patching EventEmitter. More or less any APM product and a lot other tools (e.g. zone.js - yes it's used by node.js apps not only browsers) patch this also.

Therefore you might have others between once and on.

You could rely on the single threaded nature of JS and set some inOnce flag in ContextManager and check this one instead.

[dyladan]

You could rely on the single threaded nature of JS and set some inOnce flag in ContextManager and check this one instead.

Good idea. Opened a third PR #3132

[dyladan]

I think I prefer #3133. Closing this but it can be reopened if someone disagrees.