[torchello]

Pool wait metrics currently lose the time spent by callers that cancel acquisition, while connection use time cannot distinguish BEGIN from the lifetime and termination of a transaction. This adds separate client-side lifecycle measurements and optional workload attribution so applications can identify which work holds or waits for connections.

This PR is based directly on current trunk (SQLx 0.9 / OpenTelemetry 0.33). It is independent of #41 and adds no dependencies or dependency-version changes.

Measurements

• sqlx_otel.connection.abandoned_wait.duration: records elapsed wait exactly once when a polled acquisition is dropped. Never-polled futures emit nothing; completed waits and pool timeouts retain their existing meanings.
• sqlx_otel.transaction.begin.duration: measures BEGIN after acquisition, with ok, error, and abandoned outcomes.
• sqlx_otel.transaction.duration: measures from successful BEGIN to the terminal client observation, with commit/rollback success, error, abandonment, and dropped outcomes. Its seconds-based histogram extends to one hour.

Outcomes deliberately describe what the client observed. An abandoned commit does not establish whether the server committed. Dropping a transaction queues rollback; it does not confirm completion, and can be an intentional guard-release path rather than an application error. No extra SQL, await, or SQLx ownership/Drop behavior is introduced.

Optional workload attribution

PoolAttribution::new("myapp.job.kind", "nightly_rollup").scope(future) adds one static key/value to wait, use, timeout, abandoned-wait, and transaction measurements. It is captured when acquisition first runs and stays with the lease through completion, even outside the original scope.

Attributions that reuse a connection-identity or transaction-outcome key are ignored, so they cannot overwrite canonical labels. The scope is restored after each poll, including panic unwinding. Spawned tasks require explicit PoolAttribution::inherit; pool-wide gauges/pending counts and query metrics remain untagged. Enabling a scope splits existing series, so consumers must aggregate across the added dimension to retain pool-wide totals.

Validation

• Clippy (--all-targets --all-features -- -D warnings), formatting, and strict rustdoc passed.
• Instrumented all-features suite: 604 tests passed, including SQLite, PostgreSQL, and MySQL with fresh disposable containers.
• Async-std feature run: 37 lifecycle/pool tests passed.
• Doctests: 14 passed; one pre-existing ignored example.
• Combined LLVM coverage: 99.64% lines, 99.41% regions.

Tests cover cancellation versus timeouts, never-polled futures, interleaved/nested attribution scopes, spawned work, lease transfer, and real PostgreSQL boundary errors/cancellation. A response-gated commit test verifies that the server can commit while the client records commit_abandoned.

README, rustdoc, and changelog describe lifecycle boundaries, cardinality/propagation rules, and outcome limitations.

[codecov-commenter]

⚠️ Please install the to ensure uploads and comments are reliably processed by Codecov.

Codecov Report

❌ Patch coverage is 99.60474% with 2 lines in your changes missing coverage. Please review.
✅ Project coverage is 99.81%. Comparing base (e6c3de5) to head (dee0d68).

Files with missing lines	Patch %	Lines
src/attribution.rs	99.41%	1 Missing ⚠️
src/transaction_metrics.rs	99.64%	1 Missing ⚠️
❗ Your organization needs to install the Codecov GitHub app to enable full functionality.

Additional details and impacted files

@@            Coverage Diff             @@
##            trunk      #46      +/-   ##
==========================================
- Coverage   99.85%   99.81%   -0.04%     
==========================================
  Files          15       17       +2     
  Lines        2760     3253     +493     
==========================================
+ Hits         2756     3247     +491     
- Misses          4        6       +2     

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.