go.opentelemetry.io/otel may cause fatal error out of memory allocating heap arena metadata

[LINJINTIANDE]

Describe the bug:

When running 1960322 highly actor tasks, the cpu is full, resulting in downtime.
Comment out the following lines of code, the task can run normally.

lily/tasks/actorstate/raw/actor.go

Line 24 in d451f36

_, span := otel.Tracer("").Start(ctx, "RawActorExtractor.Extract")

lily/tasks/actorstate/raw/actor.go

Line 25 in d451f36

defer span.End()

lily/tasks/actorstate/raw/actor.go

Line 26 in d451f36

if span.IsRecording() {

lily/tasks/actorstate/raw/actor.go

Line 27 in d451f36

span.SetAttributes(a.Attributes()...)

lily/tasks/actorstate/raw/actor.go

Line 27 in d451f36

span.SetAttributes(a.Attributes()...)

Steps to Reproduce:

Running a walk job like this:
lily job run --storage=lily --tasks=actor --storage=lily walk --from=1960322 --to=1960322

Lily Version: v0.12.0

[frrist]

Interesting. Can you share the command used to start the daemon? Was tracing enabled?

[LINJINTIANDE]

run lily's script