Repository navigation
[P2] Add self-observability metrics (queue length, in-flight, WS reconnects, last header) #45
Description
Priority: P2 (feature) · Effort: S
Problem
Today you can't tell from metrics alone that the verifier is falling behind or that the WebSocket keeps reconnecting.
Proposed metrics
ev_metrics_verifier_queue_length(gauge)ev_metrics_verifier_inflight_blocks(gauge)ev_metrics_ws_reconnects_total(counter)ev_metrics_last_header_received_timestamp(gauge, unix seconds)
Pros / Cons
- ✅ Lets you alert on the monitor's own health.
⚠️ 4 more series.
Acceptance criteria
- Tests in
pkg/metrics/metrics_test.go. - README updated (per CLAUDE.md).
Activity
auricom commentedon Oct 20, 2026
Scope update after #58: the verifier now uses
jobs and timer-based retries (time.AfterFunc) and backfills after reconnects. Suggested metric semantics:verifier_queue_length:len(blockQueue)verifier_inflight_blocks: blocks between first attempt and verified or exhausted, including those waiting on a retry timer (count up and down inverifyBlock)ws_reconnects_total: increment wherereconnectSubscriptionreturns a new subscriptionverifier_backfilled_blocks_totalandverifier_backfill_dropped_total(the 1000-block cap inbackfill)The verifier tests use fakes, so these can be unit-tested.