fix: slot-ticker rename, eventsync log accuracy, ProbeAll error masking

[Roy-blox]

Summary

Three independent small fixes bundled together:

• slot-ticker: rename Next->Advance to clarify assumptions #2784 — Rename SlotTicker.Next() → Advance() across interface, implementation, generated mock, and all callers (18 files). The old name was misleading — the method advances the ticker as a side-effect; Advance makes that explicit.
• eventsync: last_processed_block on ongoing-sync failure is the historical-sync block, not the failure block #2878 — eventsync: the fatal log on SyncOngoing failure was emitting last_processed_block captured from the historical sync end, not the block SyncOngoing actually started from. On a long-running node these differ by millions of blocks. Now captures from_block at goroutine-spawn time.
• hprobe: ProbeAll surfaces its own early-cancel context.Canceled in the joined error #2880 — hprobe.ProbeAll: when one component fails it calls cancel(), causing sibling goroutines to return context.Canceled from their retry-wait select. These were being errors.Join'd into the result, so errors.Is(err, context.Canceled) was true even when a real probe failure was present — masking it for callers. Now silently drops context.Canceled from siblings.

Test plan

• go test ./operator/slotticker/... ./operator/duties/... ./operator/fee_recipient/... ./doppelganger/... ./ibft/storage/... ./hprobe/... ./cli/operator/... — 167 pass
• make lint — 0 issues

🤖 Generated with Claude Code

[codecov]

Codecov Report

❌ Patch coverage is 73.91304% with 6 lines in your changes missing coverage. Please review.
✅ Project coverage is 61.3%. Comparing base (82a9f4f) to head (594ed35).

Files with missing lines	Patch %	Lines
cli/operator/eventsync.go	0.0%	3 Missing ⚠️
operator/dutytracer/collector.go	0.0%	2 Missing ⚠️
doppelganger/doppelganger.go	0.0%	1 Missing ⚠️

☔ 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.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[greptile-apps]

Greptile Summary

Three independent targeted fixes: SlotTicker.Next() is renamed to Advance() across the interface, implementation, generated mock, and all 18 call sites; eventsync now captures from_block at goroutine-spawn time for an accurate fatal-log field; and hprobe.ProbeAll filters out context.Canceled errors from sibling goroutines so a real probe failure is no longer masked.

• Rename (Next → Advance): mechanical, complete, and consistent across all callers and mocks — one pair of doc-comment lines in the SlotTicker interface block still reference the old name.
• eventsync log fix: ongoingFromBlock is now captured before the goroutine is spawned, replacing the stale last_processed_block log field that reflected the historical-sync end block rather than the SyncOngoing start block.
• ProbeAll error masking fix: sibling goroutines that return context.Canceled because a failing peer called cancel() now exit silently instead of injecting a context.Canceled entry into the joined error, ensuring callers see only genuine probe failures.

Confidence Score: 4/5

Safe to merge; all three fixes are narrow, well-scoped, and leave no behavioral regression in the changed paths.

The rename is mechanical and exhaustive, the eventsync log fix correctly captures the value before goroutine spawn, and the ProbeAll change is consistent with the pre-existing context.Canceled handling already present in probeComponent. The only gap is two doc-comment lines in the SlotTicker interface that still say Next after the rename.

operator/slotticker/slotticker.go — two interface doc-comment lines still reference the old method name Next.

Important Files Changed

Filename	Overview
operator/slotticker/slotticker.go	Renames Next() → Advance() on the SlotTicker interface and its implementation; two doc-comment references to "Next" were missed in the interface block (lines 19, 27).
operator/slotticker/mocks/slotticker.go	Generated mock updated to reflect the Next → Advance rename; looks complete and correct.
cli/operator/eventsync.go	Captures ongoingFromBlock before goroutine spawn so the fatal log always reflects the block SyncOngoing actually started from, not the historical-sync end block.
hprobe/prober.go	Filters out context.Canceled from sibling goroutines in ProbeAll so a real probe failure is not masked; fix is correct and consistent with existing handling in probeComponent.
operator/duties/attester.go	Next() → Advance() rename applied consistently (both initial call and in-loop refresh).
doppelganger/doppelganger.go	Single-site Next() → Advance() rename applied correctly.
ibft/storage/store.go	Next() → Advance() rename applied correctly in PruneContinuously loop.
protocol/v2/ssv/runner/validator_registration.go	Next() → Advance() rename applied correctly in VRSubmitter start loop.

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    PA["ProbeAll(ctx)"] --> CC["context.WithCancel(ctx)"]
    CC --> PG["Launch goroutine per component"]
    PG --> PC["probeComponent(ctx, n)"]
    PC -->|success| WD["wg.Done() — silent"]
    PC -->|context.Canceled| NEW["NEW: skip errsCh send\nwg.Done() — silent"]
    PC -->|real error| ES["errsCh ← fmt.Errorf(...)"]
    ES --> CL["cancel() — cancels siblings"]
    CL --> NEW
    WD --> WG["wg.Wait() → close(errsCh)"]
    NEW --> WG
    WG --> RL["range errsCh → errors.Join"]
    RL --> RET["return joined real errors only"]

    style NEW fill:#d4edda,stroke:#28a745
    style CL fill:#fff3cd,stroke:#ffc107

Loading

%%{init: {'theme': 'base', 'themeVariables': {"darkMode": true, "background": "#0d1117", "primaryColor": "#21262d", "primaryTextColor": "#e6edf3", "primaryBorderColor": "#8b949e", "lineColor": "#8b949e", "textColor": "#e6edf3", "edgeLabelBackground": "#161b22", "actorBkg": "#21262d", "actorBorder": "#8b949e", "actorTextColor": "#e6edf3", "actorLineColor": "#8b949e", "signalColor": "#8b949e", "signalTextColor": "#e6edf3", "noteBkgColor": "#373320", "noteBorderColor": "#d4a72c", "noteTextColor": "#f0e6c0", "labelBoxBkgColor": "#21262d", "labelBoxBorderColor": "#8b949e", "labelTextColor": "#e6edf3", "loopTextColor": "#e6edf3", "activationBkgColor": "#30363d", "activationBorderColor": "#8b949e"}}}%%
flowchart TD
    PA["ProbeAll(ctx)"] --> CC["context.WithCancel(ctx)"]
    CC --> PG["Launch goroutine per component"]
    PG --> PC["probeComponent(ctx, n)"]
    PC -->|success| WD["wg.Done() — silent"]
    PC -->|context.Canceled| NEW["NEW: skip errsCh send\nwg.Done() — silent"]
    PC -->|real error| ES["errsCh ← fmt.Errorf(...)"]
    ES --> CL["cancel() — cancels siblings"]
    CL --> NEW
    WD --> WG["wg.Wait() → close(errsCh)"]
    NEW --> WG
    WG --> RL["range errsCh → errors.Join"]
    RL --> RET["return joined real errors only"]

    style NEW fill:#d4edda,stroke:#28a745
    style CL fill:#fff3cd,stroke:#ffc107

Loading

Comments Outside Diff (1)

1. operator/slotticker/slotticker.go, line 17-28 (link)

   Two doc-comment references to the old method name were not updated alongside the rename. Both "calling Next method periodically" and "corresponds to Next" should now say Advance.

_{Reviews (1): Last reviewed commit: "fix(hprobe): drop sibling context.Cancel..." | Re-trigger Greptile}

[Roy-blox]

QA — Deploy Test Results ✅

Field	Value
Environment	stage-hoodi
Nodes	97-100 (COMMITTEE-97_98_99_100)
Commit	594ed35b45e8b5b79938698d6edab084e91b43bc
Sensor	ssv-stage-fix-small-cleanups-2784-2878-6c35a3-1

---

1. Commit Verification ✅

All 4 nodes confirmed running the PR head commit:

2026-06-16T09:39:20Z ssv-node-97  starting SSV-Node:untagged-594ed35b45e8b5b79938698d6edab084e91b43bc
2026-06-16T09:39:23Z ssv-node-98  starting SSV-Node:untagged-594ed35b45e8b5b79938698d6edab084e91b43bc
2026-06-16T09:39:25Z ssv-node-99  starting SSV-Node:untagged-594ed35b45e8b5b79938698d6edab084e91b43bc
2026-06-16T09:39:25Z ssv-node-100 starting SSV-Node:untagged-594ed35b45e8b5b79938698d6edab084e91b43bc

2. Duty Execution ✅

1000 duty-start events on node-97 in a 5-minute window — the slot ticker fires on every slot. Duties completing successfully end-to-end:

2026-06-16T09:43:36Z ssv-node-97  ✔️ successfully finished duty processing (no selection proofs)
  runner_role=SYNC_COMMITTEE_CONTRIBUTION slot=3282468
  pre_consensus_time=0.02654s total_consensus_time=0.04282s total_duty_time=0.06753s

3. QBFT Consensus ✅

Round 1 decisions observed across all duty types — no round changes, no timeouts:

runner_role=AGGREGATOR slot=3282471 qbft_round=1 qbft_height=3282471

4. Errors ✅

No application errors or panics. Only benign P2P gossip noise:

ERROR could not publish p2p message topic=62 error="validation ignored"

("validation ignored" = peer rejected a late/duplicate message — expected background noise.)

5. Metrics ✅

Operator	Peers	CL Sync	Consensus avg (COMMITTEE, 5m)	Submissions/10m (Aggregator)
97	4	synced	8.48s	6.94
98	5	synced	8.46s	6.91
99	3	synced	11.98s	6.95
100	4	synced	8.48s	7.53

Node-99 shows slightly elevated consensus latency (~12s vs ~8.5s for peers). This is within normal staging variance and predates this PR — no consensus logic was changed.

---

PR-Specific Checks

SlotTicker.Next() → Advance() (#2784) ✅

Pure rename — verified no behavioral regression:

• Zero double tick warnings across all 4 nodes — the ticker still advances slot-by-slot without skipping or double-firing
• 1000 duty-start events / 5 min — scheduler calling Advance() on every slot, full duty pipeline intact

eventsync: stale last_processed_block log (#2878) ✅

Defensive fix to error logging — normal operation path is unchanged. Nodes started cleanly, no failed syncing ongoing registry events observed. The fix only affects what's logged when SyncOngoing hits a fatal error; not testable in healthy infra by design.

hprobe: ProbeAll context.Canceled masking (#2880) ✅

Defensive fix to error handling — normal operation path is unchanged. All health probes passing; no probe-related errors logged. The fix only prevents context.Canceled from siblings being joined into the error when a real probe failure occurs; not testable in healthy infra by design.

---

Overall: PASS — all nodes healthy, duties executing, no regressions introduced.

[Roy-blox]

Follow-up on consensus duration figures above

The COMMITTEE consensus avg figures in the metrics table (8–12s) were startup noise — queried at --since 5m exactly when the nodes had just booted. With only a handful of consensus samples in the window, a few slow warmup rounds dominate the rate(sum)/rate(count) average.

Re-queried at --since 30m after nodes stabilised:

	COMMITTEE consensus avg (30m window)
PR nodes 97-100	~0.01–1.26s
Control nodes 1-4 (stage branch)	~0.85–0.93s

No consensus regression. Values are comparable to the stage branch control group. QA conclusion stands: PASS.

[momosh-ssv]

Might be worth updating this — the doc still says calling Next method periodically, but the method is now Advance. Since the whole point of the rename was making the name honest, the comment should follow.

[momosh-ssv]

Same here: corresponds to Next should now read Advance.

[momosh-ssv]

Worth considering a regression test for the masking path this guard fixes?
All current cases in prober_test.go use retryDelay: 0, so the retry-wait select is never where a sibling cancel() lands — the exact scenario this addresses isn't exercised.

A two-component test (one failing fast, one with a non-zero retryDelay) would lock the fix in.