Add block-update latency tracking, raw share counters, and shares bar UI

Log ZMQ→mining.notify latency in ckpool (patch 0005), expose a raw
reject counter (patch 0006), and surface both in the dashboard:

- Block latency card shows avg/last ms, wasted work, and block count
- SharesBar component shows session + all-time accepted/rejected with
  a Demon Slayer flame-slash animation on new shares
- Miners card info moved to MinersTable section header
- Latency stats and share counts persist across restarts via kv store
- Removed misleading pool-wide share stats from per-worker/user pages
  (ckpool doesn't expose per-user raw counts)

api/cmd/kamado-api/main.go

@@ -128,6 +128,7 @@ func main() {
 	go tailer.Run(ctx)
 	go agg.IngestBlockEvents(ctx, tailer.Events)
 	go agg.IngestAttemptEvents(ctx, tailer.Attempts)
+	go agg.IngestLatencyEvents(ctx, tailer.Latencies)
 
 	srv := &http.Server{
 		Addr:              cfg.ListenAddr,

api/internal/ckpool/types.go

@@ -26,6 +26,7 @@ type PoolStats struct {
 	SPS60        float64 `json:"sps60"`
 	Accepted     int64   `json:"accepted"`
 	Rejected     int64   `json:"rejected"`
+	RejectCount  int64   `json:"rejectcount"`
 	DSPS1        float64 `json:"dsps1"`
 	DSPS5        float64 `json:"dsps5"`
 	DSPS15       float64 `json:"dsps15"`

api/internal/logmon/tailer.go

@@ -49,12 +49,21 @@ type AttemptEvent struct {
 	RawLine   string
 }
 
+// LatencyEvent is emitted when ckpool logs the "Block update latency"
+// line from our patch 0005 — the milliseconds between ZMQ trigger and
+// mining.notify broadcast completion.
+type LatencyEvent struct {
+	SeenAt    time.Time
+	LatencyMs int64
+}
+
 // Tailer follows a log file, surviving rotation/truncation, and emits
 // parsed events. Create with New, then Run in a goroutine.
 type Tailer struct {
 	Path      string
 	Events    chan BlockEvent
 	Attempts  chan AttemptEvent
+	Latencies chan LatencyEvent
 	Log       *slog.Logger
 	PollWait  time.Duration // how long to sleep between EOF polls
 
@@ -80,6 +89,7 @@ func New(path string, log *slog.Logger) *Tailer {
 		Path:      path,
 		Events:    make(chan BlockEvent, 16),
 		Attempts:  make(chan AttemptEvent, 16),
+		Latencies: make(chan LatencyEvent, 16),
 		Log:       log,
 		PollWait:  500 * time.Millisecond,
 	}
@@ -94,6 +104,9 @@ var (
 	//   "Submitting possible block solve share diff N !"
 	//   "Possible remote block solve diff N !"
 	solveDiffRE = regexp.MustCompile(`(?:Possible|Submitting[^"]*possible).*block solve.*diff\s+([0-9eE.+-]+)`)
+	// Matches the latency line from our patch 0005:
+	//   "Block update latency: 92ms (ZMQ trigger to mining.notify broadcast)"
+	latencyRE = regexp.MustCompile(`Block update latency:\s+(\d+)ms`)
 )
 
 // Run blocks until ctx is cancelled. It opens the file, seeks to either
@@ -104,6 +117,7 @@ var (
 func (t *Tailer) Run(ctx context.Context) {
 	defer close(t.Events)
 	defer close(t.Attempts)
+	defer close(t.Latencies)
 
 	var (
 		f       *os.File
@@ -242,6 +256,15 @@ func (t *Tailer) Run(ctx context.Context) {
 }
 
 func (t *Tailer) handleLine(line string) {
+	if m := latencyRE.FindStringSubmatch(line); m != nil {
+		if ms, err := strconv.ParseInt(m[1], 10, 64); err == nil {
+			select {
+			case t.Latencies <- LatencyEvent{SeenAt: time.Now(), LatencyMs: ms}:
+			default:
+			}
+		}
+		return
+	}
 	if m := solveDiffRE.FindStringSubmatch(line); m != nil {
 		var d float64
 		if v, err := strconv.ParseFloat(m[1], 64); err == nil {

api/internal/state/aggregator.go

@@ -88,6 +88,19 @@ type Snapshot struct {
 	LastZMQEventAge   float64 `json:"last_zmq_event_age,omitempty"` // seconds; >=0
 	HasLastZMQEvent   bool    `json:"has_last_zmq_event"`
 
+	// Share counters: raw counts (1 submission = 1 share regardless of diff).
+	// Session = since ckpool started; AllTime = persisted across restarts.
+	SessionAccepted int64 `json:"session_accepted"`
+	SessionRejected int64 `json:"session_rejected"`
+	AllTimeAccepted int64 `json:"alltime_accepted"`
+	AllTimeRejected int64 `json:"alltime_rejected"`
+
+	// Block update latency diagnostics (ZMQ trigger → mining.notify).
+	LatencyCount    int64   `json:"latency_count"`
+	LatencyAvgMs    int64   `json:"latency_avg_ms"`
+	LatencyLastMs   int64   `json:"latency_last_ms"`
+	StaleWorkHashes float64 `json:"stale_work_hashes"`
+
 	// Health
 	CKPoolOK  bool   `json:"ckpool_ok"`
 	BitcoinOK bool   `json:"bitcoin_ok"`
@@ -107,6 +120,14 @@ const (
 	kvSubmitAttempts   = "block_submit_attempts"
 	kvSubmitsConfirmed = "block_submits_confirmed"
 
+	kvLatencyCount    = "latency_count"
+	kvLatencySumMs    = "latency_sum_ms"
+	kvLatencyLastMs   = "latency_last_ms"
+	kvStaleWorkHashes = "stale_work_hashes"
+
+	kvAllTimeAccepted = "alltime_accepted"
+	kvAllTimeRejected = "alltime_rejected"
+
 	// reconcileInterval is how often we sweep recent blocks looking
 	// for missing hash/reward enrichment and reorg-orphaned hashes.
 	// 60s is fast enough to recover from a transient bitcoind hiccup
@@ -192,6 +213,22 @@ type Aggregator struct {
 	// from zmqmon. Used by /healthz to flag stale subscriptions.
 	zmqEnabled       bool
 	lastZMQEventTime time.Time
+
+	// All-time share counters (raw, not diff-weighted). Accumulated
+	// using the same delta-integration pattern as cumulative_shares.
+	allTimeAccepted       int64
+	allTimeRejected       int64
+	lastPoolAcceptedRaw   int64
+	lastPoolRejectedRaw   int64
+	hasShareCountBaseline bool
+	lastShareCountSave    time.Time
+
+	// Block update latency tracking (from ckpool patch 0005).
+	// Persisted to kv store so stats accumulate across restarts.
+	latencyCount    int64   // number of observations
+	latencySumMs    int64   // sum of all latencies in ms
+	latencyLastMs   int64   // most recent observation
+	staleWorkHashes float64 // cumulative wasted hashes = sum(latency_s * hashrate_at_event)
 }
 
 func New(ck *ckpool.Client, rpc *bitcoind.RPC, interval time.Duration, log *slog.Logger) *Aggregator {
@@ -396,6 +433,36 @@ func (a *Aggregator) refresh(ctx context.Context) {
 		a.lastCumulativeSave = now
 	}
 
+	// --- raw share count tracking (1 share = 1 submission) ---
+	// pool.Shares = raw accepted count, pool.RejectCount = raw rejected count.
+	// Delta-integrate the same way as cumulative work.
+	if next.Pool != nil {
+		curAcc := next.Pool.Shares
+		curRej := next.Pool.RejectCount
+		if a.hasShareCountBaseline {
+			if curAcc >= a.lastPoolAcceptedRaw {
+				a.allTimeAccepted += curAcc - a.lastPoolAcceptedRaw
+			}
+			if curRej >= a.lastPoolRejectedRaw {
+				a.allTimeRejected += curRej - a.lastPoolRejectedRaw
+			}
+		}
+		a.lastPoolAcceptedRaw = curAcc
+		a.lastPoolRejectedRaw = curRej
+		a.hasShareCountBaseline = true
+		next.SessionAccepted = curAcc
+		next.SessionRejected = curRej
+	}
+	next.AllTimeAccepted = a.allTimeAccepted
+	next.AllTimeRejected = a.allTimeRejected
+
+	// Persist share counts at most once per minute.
+	if a.Store != nil && now.Sub(a.lastShareCountSave) >= time.Minute {
+		_ = a.Store.SetKV(kvAllTimeAccepted, strconv.FormatInt(a.allTimeAccepted, 10))
+		_ = a.Store.SetKV(kvAllTimeRejected, strconv.FormatInt(a.allTimeRejected, 10))
+		a.lastShareCountSave = now
+	}
+
 	// --- hashrate history sample (once per minute) ---
 	if now.Sub(a.lastHRSampleAt) >= time.Minute {
 		p := HashratePoint{T: now.Unix(), V: next.HashrateHs}
@@ -455,6 +522,12 @@ func (a *Aggregator) refresh(ctx context.Context) {
 		next.LastZMQEventAge = time.Since(a.lastZMQEventTime).Seconds()
 		next.HasLastZMQEvent = true
 	}
+	next.LatencyCount = a.latencyCount
+	if a.latencyCount > 0 {
+		next.LatencyAvgMs = a.latencySumMs / a.latencyCount
+	}
+	next.LatencyLastMs = a.latencyLastMs
+	next.StaleWorkHashes = a.staleWorkHashes
 	a.snap = next
 	cb := a.OnRefresh
 	pushed := next
@@ -500,6 +573,50 @@ func (a *Aggregator) loadPersistedState() {
 		}
 	}
 
+	// Restore all-time share counts.
+	if v, err := a.Store.GetKV(kvAllTimeAccepted); err == nil && v != "" {
+		if n, perr := strconv.ParseInt(v, 10, 64); perr == nil {
+			a.mu.Lock()
+			a.allTimeAccepted = n
+			a.mu.Unlock()
+		}
+	}
+	if v, err := a.Store.GetKV(kvAllTimeRejected); err == nil && v != "" {
+		if n, perr := strconv.ParseInt(v, 10, 64); perr == nil {
+			a.mu.Lock()
+			a.allTimeRejected = n
+			a.mu.Unlock()
+		}
+	}
+
+	// Restore latency stats.
+	a.mu.Lock()
+	if v, err := a.Store.GetKV(kvLatencyCount); err == nil && v != "" {
+		if n, perr := strconv.ParseInt(v, 10, 64); perr == nil {
+			a.latencyCount = n
+		}
+	}
+	if v, err := a.Store.GetKV(kvLatencySumMs); err == nil && v != "" {
+		if n, perr := strconv.ParseInt(v, 10, 64); perr == nil {
+			a.latencySumMs = n
+		}
+	}
+	if v, err := a.Store.GetKV(kvLatencyLastMs); err == nil && v != "" {
+		if n, perr := strconv.ParseInt(v, 10, 64); perr == nil {
+			a.latencyLastMs = n
+		}
+	}
+	if v, err := a.Store.GetKV(kvStaleWorkHashes); err == nil && v != "" {
+		if f, perr := strconv.ParseFloat(v, 64); perr == nil {
+			a.staleWorkHashes = f
+		}
+	}
+	if a.latencyCount > 0 {
+		a.Log.Info("latency stats loaded", "count", a.latencyCount,
+			"avg_ms", a.latencySumMs/a.latencyCount, "last_ms", a.latencyLastMs)
+	}
+	a.mu.Unlock()
+
 	cutoff := time.Now().Add(-24 * time.Hour).Unix()
 	if samples, err := a.Store.HashrateSince(cutoff); err != nil {
 		a.Log.Warn("hashrate history load failed", "err", err)

api/internal/state/blocks.go

@@ -80,6 +80,39 @@ func (a *Aggregator) IngestAttemptEvents(ctx context.Context, events <-chan logm
 	}
 }
 
+// IngestLatencyEvents reads block-update latency events from the tailer
+// and accumulates stats for the dashboard. Stats are persisted to the
+// kv store so they survive restarts.
+func (a *Aggregator) IngestLatencyEvents(ctx context.Context, events <-chan logmon.LatencyEvent) {
+	for {
+		select {
+		case <-ctx.Done():
+			return
+		case ev, ok := <-events:
+			if !ok {
+				return
+			}
+			a.mu.Lock()
+			a.latencyCount++
+			a.latencySumMs += ev.LatencyMs
+			a.latencyLastMs = ev.LatencyMs
+			// Wasted hashes = latency in seconds × current 5m hashrate.
+			hs5m := a.snap.HashrateHs5m
+			a.staleWorkHashes += (float64(ev.LatencyMs) / 1000.0) * hs5m
+			count, sum, last, stale := a.latencyCount, a.latencySumMs, a.latencyLastMs, a.staleWorkHashes
+			a.mu.Unlock()
+			a.Log.Info("logmon: block update latency", "ms", ev.LatencyMs, "count", count)
+			// Blocks are infrequent (~10min); persist every event.
+			if a.Store != nil {
+				_ = a.Store.SetKV(kvLatencyCount, strconv.FormatInt(count, 10))
+				_ = a.Store.SetKV(kvLatencySumMs, strconv.FormatInt(sum, 10))
+				_ = a.Store.SetKV(kvLatencyLastMs, strconv.FormatInt(last, 10))
+				_ = a.Store.SetKV(kvStaleWorkHashes, strconv.FormatFloat(stale, 'f', -1, 64))
+			}
+		}
+	}
+}
+
 // IngestBlockEvents reads block events from the tailer and appends them
 // to the snapshot's block history. Runs until ctx is cancelled.
 func (a *Aggregator) IngestBlockEvents(ctx context.Context, events <-chan logmon.BlockEvent) {

ckpool/patches/0005-log-zmq-to-notify-latency.patch

@@ -0,0 +1,38 @@
+diff --git a/src/stratifier.c b/src/stratifier.c
+index a039aad3..d6355d6f 100644
+--- a/src/stratifier.c
++++ b/src/stratifier.c
+@@ -429,6 +429,9 @@ struct stratifier_data {
+ 	/* Time we last sent out a stratum update */
+ 	time_t update_time;
+ 
++	/* Timestamp of last ZMQ block trigger for latency measurement */
++	tv_t zmq_trigger_time;
++
+ 	int64_t workbase_id;
+ 	int64_t blockchange_id;
+ 	int session_id;
+@@ -1519,6 +1522,15 @@ retry:
+ 		stratum_broadcast_update(sdata, wb, new_block);
+ 	ret = true;
+ 	LOGINFO("Broadcast updated stratum base");
++	if (new_block && sdata->zmq_trigger_time.tv_sec) {
++		tv_t now;
++		tv_time(&now);
++		LOGWARNING("Block update latency: %dms (ZMQ trigger to mining.notify broadcast)",
++			   ms_tvdiff(&now, &sdata->zmq_trigger_time));
++	}
++	/* Always clear ZMQ trigger so a stale timestamp from a reconnection
++	 * or duplicate detection does not contaminate the next measurement. */
++	sdata->zmq_trigger_time.tv_sec = 0;
+ 	/* Update transactions after stratum broadcast to not delay
+ 	 * propagation. */
+ 	if (likely(txns))
+@@ -8544,6 +8556,7 @@ static void *zmqnotify(void *arg)
+ 					LOGDEBUG("ZMQ sequence number");
+ 					break;
+ 				case 32:
++					tv_time(&sdata->zmq_trigger_time);
+ 					update_base(sdata, GEN_PRIORITY);
+ 					__bin2hex(hexhash, zmq_msg_data(&message), 32);
+ 					LOGNOTICE("ZMQ block hash %s", hexhash);

ckpool/patches/0006-expose-raw-reject-count-in-poolstats.patch

@@ -0,0 +1,69 @@
+diff --git a/src/stratifier.c b/src/stratifier.c
+index d6355d6f..09b68097 100644
+--- a/src/stratifier.c
++++ b/src/stratifier.c
+@@ -71,6 +71,8 @@ struct pool_stats {
+ 	int64_t accounted_diff_shares;
+ 	int64_t unaccounted_rejects;
+ 	int64_t accounted_rejects;
++	int64_t unaccounted_reject_count;
++	int64_t accounted_reject_count;
+ 
+ 	/* Diff shares per second for 1/5/15... minute rolling averages */
+ 	double dsps1;
+@@ -4508,13 +4510,13 @@ static void get_poolstats(sdata_t *sdata, int *sockd)
+ 	json_t *val;
+ 
+ 	mutex_lock(&sdata->stats_lock);
+-	JSON_CPACK(val, "{si,si,si,si,si,sI,sf,sf,sf,sf,sI,sI,sf,sf,sf,sf,sf,sf,sf}",
++	JSON_CPACK(val, "{si,si,si,si,si,sI,sf,sf,sf,sf,sI,sI,sI,sf,sf,sf,sf,sf,sf,sf}",
+ 		   "start", stats->start_time.tv_sec, "update", stats->last_update.tv_sec,
+ 	    "workers", stats->workers + stats->remote_workers, "users", stats->users + stats->remote_users,
+ 	    "disconnected", stats->disconnected,
+ 	    "shares", stats->accounted_shares, "sps1", stats->sps1, "sps5", stats->sps5,
+ 	    "sps15", stats->sps15, "sps60", stats->sps60, "accepted", stats->accounted_diff_shares,
+-	    "rejected", stats->accounted_rejects, "dsps1", stats->dsps1, "dsps5", stats->dsps5,
++	    "rejected", stats->accounted_rejects, "rejectcount", stats->accounted_reject_count, "dsps1", stats->dsps1, "dsps5", stats->dsps5,
+ 	    "dsps15", stats->dsps15, "dsps60", stats->dsps60, "dsps360", stats->dsps360,
+ 	    "dsps1440", stats->dsps1440, "dsps10080", stats->dsps10080);
+ 	mutex_unlock(&sdata->stats_lock);
+@@ -5625,8 +5627,10 @@ static void add_submit(ckpool_t *ckp, stratum_instance_t *client, const double d
+ 	if (valid) {
+ 		ckp_sdata->stats.unaccounted_shares++;
+ 		ckp_sdata->stats.unaccounted_diff_shares += diff;
+-	} else
++	} else {
+ 		ckp_sdata->stats.unaccounted_rejects += diff;
++		ckp_sdata->stats.unaccounted_reject_count++;
++	}
+ 	mutex_unlock(&ckp_sdata->uastats_lock);
+ 
+ 	/* Count only accepted and stale rejects in diff calculation. */
+@@ -8325,7 +8329,7 @@ out_status:
+ 		for (i = 0; i < 32; i++) {
+ 			int64_t unaccounted_shares,
+ 				unaccounted_diff_shares,
+-				unaccounted_rejects;
++				unaccounted_rejects, unaccounted_reject_count;
+ 
+ 			ts_to_tv(&diff, &stats->last_update);
+ 			cksleep_ms_r(&stats->last_update, 1875);
+@@ -8339,15 +8343,18 @@ out_status:
+ 			unaccounted_shares = stats->unaccounted_shares;
+ 			unaccounted_diff_shares = stats->unaccounted_diff_shares;
+ 			unaccounted_rejects = stats->unaccounted_rejects;
++			unaccounted_reject_count = stats->unaccounted_reject_count;
+ 			stats->unaccounted_shares =
+ 			stats->unaccounted_diff_shares =
+ 			stats->unaccounted_rejects = 0;
++			stats->unaccounted_reject_count = 0;
+ 			mutex_unlock(&sdata->uastats_lock);
+ 
+ 			mutex_lock(&sdata->stats_lock);
+ 			stats->accounted_shares += unaccounted_shares;
+ 			stats->accounted_diff_shares += unaccounted_diff_shares;
+ 			stats->accounted_rejects += unaccounted_rejects;
++			stats->accounted_reject_count += unaccounted_reject_count;
+ 
+ 			decay_time(&stats->sps1, unaccounted_shares, per_tdiff, MIN1);
+ 			decay_time(&stats->sps5, unaccounted_shares, per_tdiff, MIN5);

ui/src/App.svelte

@@ -11,6 +11,7 @@
   import BestShares from "./lib/BestShares.svelte";
   import UserDetailPage from "./lib/UserDetailPage.svelte";
   import WorkerDetailPage from "./lib/WorkerDetailPage.svelte";
+  import SharesBar from "./lib/SharesBar.svelte";
   import BlockFoundAnimation from "./lib/BlockFoundAnimation.svelte";
 
   onMount(() => {
@@ -39,6 +40,7 @@
     <UserDetailPage />
   {:else}
     <PoolOverview />
+    <SharesBar />
     <HashrateChart />
     <BlocksTable />
     <BestShares />

ui/src/lib/MinersTable.svelte

@@ -96,7 +96,14 @@
 </script>
 
 <section class="card">
+  <div class="section-head">
     <h2>Miners</h2>
+    <span class="head-meta">
+      {rows.filter(r => r.online).length} online ·
+      {rows.length} workers ·
+      {new Set(rows.map(r => r.btcAddress)).size} users
+    </span>
+  </div>
   {#if rows.length === 0}
     <div class="empty">No miners connected.</div>
   {:else}
@@ -165,11 +172,21 @@
 </section>
 
 <style>
+  .section-head {
+    display: flex;
+    align-items: baseline;
+    gap: 0.75em;
+    margin-bottom: 1rem;
+  }
   h2 {
-    margin: 0 0 1rem;
+    margin: 0;
     font-size: 1.05rem;
     font-weight: 600;
   }
+  .head-meta {
+    color: var(--fg-dim);
+    font-size: 0.85em;
+  }
   .table-wrap {
     overflow-x: auto;
   }

ui/src/lib/PoolOverview.svelte

@@ -11,8 +11,6 @@
   } from "../format";
 
   const data = $derived(snap.data!);
-  const miners = $derived(data.clients?.length ?? 0);
-  const workerCount = $derived(data.workers?.length ?? 0);
   const height = $derived(data.chain?.blocks ?? 0);
 
   const poolShare = $derived.by(() => {
@@ -56,6 +54,12 @@
 
   const totalHashes = $derived(data.cumulative_shares * 2 ** 32);
 
+  // Block update latency (ZMQ → mining.notify)
+  const latencyAvg = $derived(data.latency_avg_ms ?? 0);
+  const latencyLast = $derived(data.latency_last_ms ?? 0);
+  const latencyCount = $derived(data.latency_count ?? 0);
+  const staleWork = $derived(data.stale_work_hashes ?? 0);
+
   // Luck: best_share / cumulative_work * 100%.
   // 100% = exactly expected, >100% = lucky, <100% = unlucky.
   const luckPct = $derived.by(() => {
@@ -109,10 +113,22 @@
     </div>
   </div>
 
-  <div class="card">
+  <div
-    <div class="stat-label">Miners</div>
+    class="card"
-    <div class="stat-value">{miners}</div>
+    title="Time from receiving a new block via ZMQ to pushing mining.notify to all miners. Wasted work = hashes spent mining the old block during this window."
-    <div class="stat-sub">{workerCount} workers · {data.users?.length ?? 0} users</div>
+  >
+    <div class="stat-label">Block latency</div>
+    {#if latencyCount > 0}
+      <div class="stat-value">{latencyAvg}<span class="unit">ms</span></div>
+      <div class="stat-sub">
+        last {latencyLast}ms ·
+        <span class="wasted">{formatWork(staleWork)} wasted</span>
+        <span class="stat-dim">({latencyCount} blocks)</span>
+      </div>
+    {:else}
+      <div class="stat-value">—</div>
+      <div class="stat-sub">waiting for the next block</div>
+    {/if}
   </div>
 
   <div class="card">
@@ -171,6 +187,7 @@
       <span>{retarget.remaining} blocks · ~{retarget.eta}</span>
     </div>
   </div>
+
 </section>
 
 <style>
@@ -209,6 +226,13 @@
     font-weight: 400;
     margin-left: 0.15em;
   }
+  .stat-dim {
+    color: var(--fg-dim);
+    font-size: 0.9em;
+  }
+  .wasted {
+    color: var(--bad);
+  }
   .adj.up {
     color: var(--good);
   }

ui/src/lib/SharesBar.svelte

@@ -0,0 +1,196 @@
+<script lang="ts">
+  import { untrack } from "svelte";
+  import { snap } from "../stores/snapshot.svelte";
+
+  const data = $derived(snap.data!);
+
+  const sessionAcc = $derived(data.session_accepted ?? 0);
+  const sessionRej = $derived(data.session_rejected ?? 0);
+  const allAcc = $derived(data.alltime_accepted ?? 0);
+  const allRej = $derived(data.alltime_rejected ?? 0);
+
+  const sessionTotal = $derived(sessionAcc + sessionRej);
+  const sessionRejectPct = $derived(
+    sessionTotal > 0 ? (sessionRej / sessionTotal) * 100 : 0,
+  );
+  const allTotal = $derived(allAcc + allRej);
+  const allRejectPct = $derived(
+    allTotal > 0 ? (allRej / allTotal) * 100 : 0,
+  );
+
+  // Detect new accepted shares to trigger animation
+  let prevAccepted = $state(0);
+  let pulse = $state(false);
+  $effect(() => {
+    const cur = sessionAcc;
+    const prev = untrack(() => prevAccepted);
+    if (cur > 0 && prev > 0 && cur > prev) {
+      pulse = true;
+      setTimeout(() => { pulse = false; }, 700);
+    }
+    prevAccepted = cur;
+  });
+
+  function fmtCount(n: number): string {
+    if (n >= 1e6) return (n / 1e6).toFixed(1) + "M";
+    if (n >= 1e3) return (n / 1e3).toFixed(1) + "K";
+    return n.toLocaleString();
+  }
+</script>
+
+<section class="shares-bar" class:pulse>
+  <div class="slash-effect" aria-hidden="true"></div>
+  <div class="content">
+    <h3 class="title">Shares</h3>
+    <div class="stats">
+      <div class="group">
+        <span class="group-label">Session</span>
+        <span class="accepted">{fmtCount(sessionAcc)}</span>
+        <span class="sep">/</span>
+        <span class="rejected">{fmtCount(sessionRej)}</span>
+        {#if sessionTotal > 0}
+          <span class="pct" class:warn={sessionRejectPct > 1}>
+            {sessionRejectPct.toFixed(2)}% rejected
+          </span>
+        {/if}
+      </div>
+      <div class="divider"></div>
+      <div class="group">
+        <span class="group-label">All-time</span>
+        <span class="accepted">{fmtCount(allAcc)}</span>
+        <span class="sep">/</span>
+        <span class="rejected">{fmtCount(allRej)}</span>
+        {#if allTotal > 0}
+          <span class="pct" class:warn={allRejectPct > 1}>
+            {allRejectPct.toFixed(2)}% rejected
+          </span>
+        {/if}
+      </div>
+    </div>
+  </div>
+</section>
+
+<style>
+  .shares-bar {
+    position: relative;
+    overflow: hidden;
+    border: 1px solid var(--border);
+    border-radius: 10px;
+    background: var(--bg-card);
+    padding: 0.85rem 1.25rem;
+  }
+  .content {
+    position: relative;
+    z-index: 1;
+    display: flex;
+    align-items: center;
+    justify-content: center;
+    gap: 1.5rem;
+  }
+  .title {
+    margin: 0;
+    font-size: 0.9rem;
+    font-weight: 600;
+    text-transform: uppercase;
+    letter-spacing: 0.04em;
+    color: var(--fg);
+  }
+  .stats {
+    display: flex;
+    align-items: center;
+    gap: 1.5rem;
+    flex-wrap: wrap;
+  }
+  .group {
+    display: flex;
+    align-items: baseline;
+    gap: 0.4em;
+  }
+  .group-label {
+    font-size: 0.75rem;
+    text-transform: uppercase;
+    letter-spacing: 0.04em;
+    color: var(--fg-dim);
+    margin-right: 0.3em;
+  }
+  .accepted {
+    color: var(--good);
+    font-weight: 600;
+    font-size: 1.1rem;
+    font-variant-numeric: tabular-nums;
+  }
+  .rejected {
+    color: var(--bad);
+    font-weight: 600;
+    font-size: 1.1rem;
+    font-variant-numeric: tabular-nums;
+  }
+  .sep {
+    color: var(--fg-dim);
+    font-size: 0.9rem;
+  }
+  .pct {
+    font-size: 1.1rem;
+    color: var(--fg-dim);
+    margin-left: 0.3em;
+  }
+  .pct.warn {
+    color: var(--bad);
+  }
+  .divider {
+    width: 1px;
+    height: 1.6rem;
+    background: var(--border);
+  }
+
+  /* Demon Slayer flame-slash animation on new accepted share */
+  .slash-effect {
+    position: absolute;
+    inset: 0;
+    pointer-events: none;
+    opacity: 0;
+    background: linear-gradient(
+      90deg,
+      transparent 0%,
+      rgba(255, 122, 58, 0.0) 30%,
+      rgba(255, 122, 58, 0.15) 48%,
+      rgba(255, 200, 50, 0.25) 50%,
+      rgba(255, 122, 58, 0.15) 52%,
+      rgba(255, 122, 58, 0.0) 70%,
+      transparent 100%
+    );
+  }
+  .shares-bar.pulse .slash-effect {
+    animation: flame-slash 0.7s ease-out;
+  }
+  @keyframes flame-slash {
+    0% {
+      opacity: 1;
+      transform: translateX(-100%) scaleY(1);
+    }
+    40% {
+      opacity: 1;
+      transform: translateX(0%) scaleY(1.2);
+    }
+    100% {
+      opacity: 0;
+      transform: translateX(100%) scaleY(1);
+    }
+  }
+  .shares-bar.pulse {
+    border-color: rgba(255, 122, 58, 0.5);
+    box-shadow: 0 0 12px rgba(255, 122, 58, 0.2);
+  }
+
+  @media (max-width: 600px) {
+    .content {
+      flex-direction: column;
+      align-items: flex-start;
+      gap: 0.5rem;
+    }
+    .divider {
+      width: 100%;
+      height: 1px;
+    }
+  }
+</style>

ui/src/lib/UserDetailPage.svelte

@@ -134,6 +134,7 @@
     };
   });
 
+
   const explorerBase = $derived(
     explorerBaseFor(snap.data?.chain?.chain, snap.data?.mempool_base_url),
   );
@@ -198,6 +199,7 @@
     </div>
   </section>
 
+
   <section class="card">
     <h3>Workers</h3>
     {#if workerRows.length === 0}

ui/src/lib/WorkerDetailPage.svelte

@@ -62,6 +62,7 @@
     return (workerShares / poolShares) * 100;
   });
 
+
   function onKey(ev: KeyboardEvent): void {
     if (ev.key === "Escape") clearSelection();
   }

ui/src/types.ts

@@ -10,6 +10,7 @@ export type PoolStats = {
   shares: number;
   accepted: number;
   rejected: number;
+  rejectcount: number;
   dsps1: number;
   dsps5: number;
   dsps15: number;
@@ -140,6 +141,18 @@ export type Snapshot = {
   zmq_enabled: boolean;
   has_last_zmq_event: boolean;
   last_zmq_event_age?: number; // seconds
+  // Share counters (raw, 1 submission = 1 share).
+  session_accepted: number;
+  session_rejected: number;
+  alltime_accepted: number;
+  alltime_rejected: number;
+
+  // Block update latency diagnostics (ZMQ → mining.notify).
+  latency_count: number;
+  latency_avg_ms: number;
+  latency_last_ms: number;
+  stale_work_hashes: number;
+
   ckpool_ok: boolean;
   bitcoin_ok: boolean;
   last_error?: string;