Is your feature request related to a problem? Please describe.
Debugging write contention and hot-spot keys in the MetaDB Datastore layer is difficult. When clients write to the same key concurrently, Datastore returns ABORTED errors due to pessimistic lock contention. There is no observable signal no metrics, no slow-operation logs to distinguish contention from other failure modes or to identify which store/record keys are causing elevated latency.
Describe the solution you'd like
Add OpenTelemetry instrumentation to all RunInTransaction call sites in MetaDB
Is your feature request related to a problem? Please describe.
Debugging write contention and hot-spot keys in the MetaDB Datastore layer is difficult. When clients write to the same key concurrently, Datastore returns ABORTED errors due to pessimistic lock contention. There is no observable signal no metrics, no slow-operation logs to distinguish contention from other failure modes or to identify which store/record keys are causing elevated latency.
Describe the solution you'd like
Add OpenTelemetry instrumentation to all RunInTransaction call sites in MetaDB