feat(metrics): split aether_decode_errors_total by reason

[0xfandom]

Summary

• Replace single IntCounter with IntCounterVec labelled by reason
• decode_log now returns Result<PoolEvent, DecodeReason> so the reason propagates end-to-end
• Three reason labels: unknown_topic (benign, high-volume in discovery mode), malformed_payload (real data-integrity bug — page on spikes), insufficient_topics (upstream producer bug)
• Per-reason counter values asserted in unit tests, not just presence

Files Changed

File	Change
crates/ingestion/src/event_decoder.rs	Add DecodeReason enum + as_str(); change decode_log + all inner decoders to return Result<PoolEvent, DecodeReason>; add 3 new reason-label tests (malformed, unknown, reason label contract)
crates/grpc-server/src/metrics.rs	IntCounter → IntCounterVec with reason label; inc_decode_errors(&str); render test asserts all three label series
crates/grpc-server/src/provider.rs	record_decode_failure now takes DecodeReason; trace log includes reason field; two new end-to-end tests (malformed_payload, insufficient_topics) in addition to the updated unknown_topic test
crates/integration-tests/tests/anvil_fork_test.rs	Migrate call sites from Option to Result

Acceptance Criteria

• aether_decode_errors_total{reason="..."} emits at least the three labels (unknown_topic, malformed_payload, insufficient_topics)
• Each call site in the decoder picks the correct label — asserted at decoder level (DecodeReason unit tests) AND at the provider level (end-to-end process_logs tests)
• Per-reason counter values asserted in tests (aether_decode_errors_total{reason="malformed_payload"} 1), not just metric presence
• cargo test --workspace --release + cargo clippy --workspace --release --all-targets -- -D warnings clean on touched crates
• Grafana panel updated to stack by reason — tracked separately when dashboards PR lands (per issue spec)

Test plan

• cargo build -p aether-ingestion -p aether-grpc-server --release — clean
• cargo test --workspace --release — all tests pass (72 ingestion, 64 grpc-server, rest of workspace)
• cargo clippy --workspace --release --all-targets -- -D warnings — no warnings
• New tests: test_decode_sync_malformed_payload, test_decode_v3_swap_malformed_payload, test_decode_reason_label_strings, test_process_logs_malformed_payload_reason_label, test_process_logs_insufficient_topics_reason_label
• Updated tests: test_process_logs_decode_failure_increments_counter now asserts the reason="unknown_topic" series specifically

Closes #92

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
aether	Ready	Preview, Comment	Apr 23, 2026 10:08am
aether-63xv	Ready	Preview, Comment	Apr 23, 2026 10:08am