chrono_forge-dashboard 0.2.0 → 0.3.0

data/app/jobs/chrono_forge/dashboard/bulk_retry_job.rb

@@ -0,0 +1,24 @@
+module ChronoForge
+  module Dashboard
+    # Fans out per-workflow retries off the request thread. The "Retry blocked"
+    # buttons enqueue one of these instead of looping in the controller, so the
+    # HTTP request stays fast even when thousands of workflows are blocked.
+    class BulkRetryJob < ActiveJob::Base
+      # Both failed and stalled workflows are retryable (matching the per-workflow
+      # Retry, which uses `retryable?`).
+      RETRYABLE_STATES = %i[failed stalled].map { |s| ChronoForge::Workflow.states[s] }.freeze
+
+      # The blocked workflows a run would retry: all of them, or just one branch's
+      # spawned children. Exposed so the controller can report a count up front.
+      def self.retryable(branch_log = nil)
+        base = branch_log ? branch_log.spawned_workflows : ChronoForge::Workflow.all
+        base.where(state: RETRYABLE_STATES)
+      end
+
+      def perform(branch_log_id = nil)
+        branch_log = branch_log_id && ChronoForge::ExecutionLog.find(branch_log_id)
+        self.class.retryable(branch_log).find_each(&:retry_later)
+      end
+    end
+  end
+end

data/app/queries/chrono_forge/dashboard/overview_query.rb

@@ -0,0 +1,62 @@
+module ChronoForge
+  module Dashboard
+    # Fleet-wide throughput: how much each workflow class has processed, plus its
+    # current in-flight and blocked load. One GROUP BY (job_class, state) scan
+    # feeds the whole page — the honest cost of a "totals across everything" view
+    # (a total can't be capped the way StatsQuery caps the hot workflow list).
+    class OverviewQuery
+      IN_FLIGHT = %i[idle running].map { |s| ChronoForge::Workflow.states[s] }.freeze
+      BLOCKED = %i[failed stalled].map { |s| ChronoForge::Workflow.states[s] }.freeze
+
+      # A class's counts, bucketed the way the Overview reads them: processed
+      # (done), in-flight (live work), blocked (needs triage).
+      Row = Struct.new(:job_class, :processed, :in_flight, :blocked) do
+        def total = processed + in_flight + blocked
+      end
+
+      # Fleet-wide card totals — each an independent COUNT so its turbo-frame can
+      # load without paying for the per-class GROUP BY the table frame runs.
+      def self.processed_total = ChronoForge::Workflow.completed.count
+
+      def self.in_flight_total = ChronoForge::Workflow.where(state: IN_FLIGHT).count
+
+      def self.blocked_total = ChronoForge::Workflow.where(state: BLOCKED).count
+
+      def rows
+        @rows ||= build_rows
+      end
+
+      # Synthetic bottom row summing every class.
+      def totals
+        @totals ||= Row.new(
+          job_class: nil,
+          processed: rows.sum(&:processed),
+          in_flight: rows.sum(&:in_flight),
+          blocked: rows.sum(&:blocked)
+        )
+      end
+
+      private
+
+      def build_rows
+        grouped = ChronoForge::Workflow.group(:job_class, :state).count
+        # enum key normalization: group(:state).count may key by the raw integer
+        # (older Rails) or the label (newer) — invert states so either resolves.
+        state_name = ChronoForge::Workflow.states.invert
+        by_class = Hash.new { |h, k| h[k] = Hash.new(0) }
+        grouped.each do |(klass, raw_state), n|
+          by_class[klass][state_name[raw_state] || raw_state.to_s] += n
+        end
+
+        by_class.map { |klass, counts|
+          Row.new(
+            job_class: klass,
+            processed: counts["completed"],
+            in_flight: counts["idle"] + counts["running"],
+            blocked: counts["failed"] + counts["stalled"]
+          )
+        }.sort_by { |r| [-r.processed, r.job_class.to_s] }
+      end
+    end
+  end
+end

data/app/queries/chrono_forge/dashboard/workflows_query.rb

@@ -73,13 +73,18 @@ module ChronoForge
       end
 
       # "blocked" is a virtual filter (failed + stalled) used by the branch
-      # children triage view to default to the actionable subset.
+      # children triage view to default to the actionable subset. "in_flight"
+      # (idle + running) is its live-work counterpart, drilled into from the
+      # Overview's in-flight column.
       BLOCKED_STATES = %i[failed stalled].map { |s| ChronoForge::Workflow.states[s] }.freeze
+      IN_FLIGHT_STATES = %i[idle running].map { |s| ChronoForge::Workflow.states[s] }.freeze
 
       def filtered
         s = @base
         if @state == "blocked"
           s = s.where(state: BLOCKED_STATES)
+        elsif @state == "in_flight"
+          s = s.where(state: IN_FLIGHT_STATES)
         elsif @state && ChronoForge::Workflow.states.key?(@state)
           s = s.where(state: ChronoForge::Workflow.states[@state])
         end

data/app/views/chrono_forge/dashboard/analytics/index.html.erb

@@ -17,15 +17,26 @@
   <% end %>
 </div>
 
+<% trend = cf_window_trend(@buckets) %>
 <div class="mb-5 grid grid-cols-2 gap-3 sm:grid-cols-4">
-  <div class="cf-card p-4">
+  <div class="cf-card flex flex-col p-4">
     <div class="text-xs uppercase tracking-wide text-zinc-400">Completion rate</div>
     <div class="mt-1 font-mono text-2xl font-semibold text-emerald-600"><%= cf_pct(@totals[:completion_rate]) %></div>
+    <% if trend %>
+      <div class="mt-auto pt-1 text-xs <%= trend[:completion] > 0 ? "text-emerald-600" : trend[:completion] < 0 ? "text-rose-600" : "text-zinc-400" %>">
+        <%= cf_trend_arrow(trend[:completion]) %> <%= cf_trend_points(trend[:completion]) %> <span class="text-zinc-400">vs first half</span>
+      </div>
+    <% end %>
   </div>
-  <div class="cf-card p-4">
+  <div class="cf-card flex flex-col p-4">
     <div class="text-xs uppercase tracking-wide text-zinc-400">Workflow failure rate</div>
     <div class="mt-1 font-mono text-2xl font-semibold <%= (@totals[:failure_rate] || 0) > 0 ? "text-rose-600" : "text-zinc-900" %>"><%= cf_pct(@totals[:failure_rate]) %></div>
     <div class="mt-0.5 text-xs text-zinc-400"><%= number_with_delimiter(@totals[:failed]) %> failed</div>
+    <% if trend && trend[:failure] != 0 %>
+      <div class="mt-auto pt-1 text-xs <%= trend[:failure] > 0 ? "text-rose-600" : "text-emerald-600" %>">
+        <%= cf_trend_arrow(trend[:failure]) %> <%= cf_trend_points(trend[:failure]) %> <span class="text-zinc-400">vs first half</span>
+      </div>
+    <% end %>
   </div>
   <div class="cf-card p-4">
     <div class="text-xs uppercase tracking-wide text-zinc-400">Avg duration</div>
@@ -79,24 +90,32 @@
 <% end %>
 
 <section class="cf-card p-5">
-  <h2 class="mb-4 text-xs font-medium uppercase tracking-wide text-zinc-500">Daily throughput</h2>
+  <h2 class="mb-4 text-xs font-medium uppercase tracking-wide text-zinc-500">Daily throughput <span class="normal-case text-zinc-400">— volume &amp; health</span></h2>
   <% if @buckets.any? %>
     <% max = @buckets.map(&:terminal).max.to_f %>
-    <div class="space-y-2">
+    <div class="space-y-1">
       <% @buckets.reverse_each do |b| %>
-        <div class="flex items-center gap-3 text-sm">
-          <div class="w-24 shrink-0 font-mono text-xs text-zinc-500"><%= b.day %></div>
+        <% flagged = cf_day_flagged?(b) %>
+        <% rate = cf_day_rate(b) %>
+        <div class="-mx-2 flex items-center gap-3 rounded px-2 py-1 text-sm <%= "bg-amber-50" if flagged %>">
+          <div class="w-24 shrink-0 font-mono text-xs <%= flagged ? "text-amber-700" : "text-zinc-500" %>"><%= b.day %></div>
           <div class="flex h-4 min-w-0 flex-1 overflow-hidden rounded bg-zinc-100">
             <div class="cf-bar bg-emerald-400 <%= cf_bar_width(b.completed, max) %>"></div>
             <div class="cf-bar bg-rose-400 <%= cf_bar_width(b.failed, max) %>"></div>
           </div>
-          <div class="w-32 shrink-0 text-right font-mono text-xs tabular-nums text-zinc-500">
+          <div class="w-28 shrink-0 text-right font-mono text-xs tabular-nums text-zinc-500">
             <span class="text-emerald-600"><%= number_with_delimiter(b.completed) %></span>
             <% if b.failed > 0 %><span class="text-rose-600"> / <%= number_with_delimiter(b.failed) %></span><% end %>
           </div>
+          <div class="w-16 shrink-0 text-right">
+            <% if rate %>
+              <span class="cf-pill <%= flagged ? "cf-pill-stalled" : "cf-pill-completed" %>" title="<%= number_with_delimiter(b.completed) %> of <%= number_with_delimiter(b.terminal) %> terminal workflows completed"><%= cf_pct(rate) %><%= " ⚠" if flagged %></span>
+            <% end %>
+          </div>
         </div>
       <% end %>
     </div>
+    <p class="mt-3 text-xs text-zinc-400">Completion rate per day; a row is flagged when at least <%= (ChronoForge::Dashboard::DashboardHelper::DAY_FAILURE_FLAG * 100).to_i %>% of the day's terminal workflows failed.</p>
   <% else %>
     <p class="py-8 text-center text-sm text-zinc-500">No workflows completed or failed in this window.</p>
   <% end %>

data/app/views/chrono_forge/dashboard/branch_children/show.html.erb

@@ -5,8 +5,11 @@
     <h1 class="text-lg font-semibold tracking-tight">Branch <span class="break-all font-mono"><%= @branch.name %></span></h1>
     <p class="mt-1 break-all text-sm text-zinc-500">children of <span class="font-mono"><%= @workflow.key %></span></p>
   </div>
+  <%# Disable + relabel while submitting, and survive the poll morph — see the
+      matching button on the workflows index for the full rationale. %>
   <%= button_to "Retry all blocked", bulk_retry_workflow_branch_path(@workflow, @branch_log), method: :post,
-        form: {data: {confirm: "Re-enqueue every failed/stalled child of this branch?"}}, class: "cf-btn cf-btn-primary shrink-0" %>
+        form: {data: {confirm: "Re-enqueue every failed/stalled child of this branch?", cf_poll_preserve: true}},
+        data: {turbo_submits_with: "Retrying…"}, class: "cf-btn cf-btn-primary shrink-0" %>
 </div>
 
 <%# Live poll stats from the branch log's metadata (throughput/ETA while draining,
@@ -74,19 +77,21 @@
 
   <div class="cf-card overflow-hidden">
     <div class="overflow-x-auto">
-      <table class="w-full min-w-[40rem] text-sm">
+      <table class="w-full min-w-[48rem] text-sm">
         <thead>
           <tr class="border-b border-zinc-200 bg-zinc-50 text-left text-xs uppercase tracking-wide text-zinc-500">
             <th class="px-4 py-2.5 font-medium">Class</th>
             <th class="px-4 py-2.5 font-medium">Key</th>
             <th class="px-4 py-2.5 font-medium">State</th>
+            <th class="px-4 py-2.5 text-right font-medium">Duration</th>
             <th class="px-4 py-2.5 text-right font-medium">Started</th>
             <th class="px-4 py-2.5 text-right font-medium">Next run</th>
             <th class="px-4 py-2.5 text-right font-medium">Updated</th>
           </tr>
         </thead>
         <tbody class="divide-y divide-zinc-100">
-          <%= render partial: "chrono_forge/dashboard/workflows/workflow_row", collection: @children, as: :workflow, locals: {waits: @waits} %>
+          <% max_dur = @children.filter_map { |w| cf_row_duration_secs(w) }.max || 1 %>
+          <%= render partial: "chrono_forge/dashboard/workflows/workflow_row", collection: @children, as: :workflow, locals: {waits: @waits, max_dur: max_dur} %>
         </tbody>
       </table>
     </div>

data/app/views/chrono_forge/dashboard/overview/_card_skeleton.html.erb

@@ -0,0 +1,5 @@
+<div class="cf-card p-4">
+  <div class="text-xs uppercase tracking-wide text-zinc-400"><%= label %></div>
+  <div class="mt-2 h-7 w-20 animate-pulse rounded bg-zinc-100"></div>
+  <div class="mt-2 h-3 w-24 animate-pulse rounded bg-zinc-100"></div>
+</div>

data/app/views/chrono_forge/dashboard/overview/_stat_card.html.erb

@@ -0,0 +1,5 @@
+<%= link_to href, class: "cf-card block p-4 hover:border-zinc-300" do %>
+  <div class="text-xs uppercase tracking-wide text-zinc-400"><%= label %></div>
+  <div class="mt-1 font-mono text-2xl font-semibold tabular-nums <%= color %>"><%= number_with_delimiter(count) %></div>
+  <div class="mt-0.5 text-xs text-zinc-400"><%= sub %></div>
+<% end %>

data/app/views/chrono_forge/dashboard/overview/blocked.html.erb

@@ -0,0 +1,5 @@
+<turbo-frame id="cf-ov-blocked">
+  <%= render "stat_card", href: workflows_path(state: "blocked", hide_branches: "0"),
+        label: "Blocked", count: @count, sub: "failed + stalled",
+        color: (@count > 0) ? "text-rose-600" : "text-zinc-900" %>
+</turbo-frame>

data/app/views/chrono_forge/dashboard/overview/classes.html.erb

@@ -0,0 +1,65 @@
+<% totals = @overview.totals %>
+<turbo-frame id="cf-ov-classes">
+  <div class="cf-card overflow-hidden">
+    <div class="overflow-x-auto">
+      <table class="w-full min-w-[46rem] text-sm">
+        <thead>
+          <tr class="border-b border-zinc-200 bg-zinc-50 text-left text-xs uppercase tracking-wide text-zinc-500">
+            <th class="px-4 py-2.5 font-medium">Class</th>
+            <th class="px-4 py-2.5 text-right font-medium">Processed</th>
+            <th class="px-4 py-2.5 font-medium">Share of processed</th>
+            <th class="px-4 py-2.5 text-right font-medium">In flight</th>
+            <th class="px-4 py-2.5 text-right font-medium">Blocked</th>
+          </tr>
+        </thead>
+        <tbody class="divide-y divide-zinc-100">
+          <% @overview.rows.each do |row| %>
+            <tr class="hover:bg-zinc-50">
+              <td class="px-4 py-2.5">
+                <%= link_to row.job_class, analytics_path(class: row.job_class), class: "font-mono text-zinc-900 hover:underline" %>
+              </td>
+              <td class="px-4 py-2.5 text-right font-mono tabular-nums">
+                <%= link_to number_with_delimiter(row.processed), workflows_path(job_class: row.job_class, state: "completed", hide_branches: "0"), class: "text-zinc-900 hover:underline" %>
+              </td>
+              <td class="px-4 py-2.5">
+                <% share = totals.processed.zero? ? 0 : (row.processed * 100.0 / totals.processed) %>
+                <div class="flex items-center gap-2">
+                  <%= cf_meter(cf_bar_width(row.processed, totals.processed), "bg-zinc-400", track: "w-24", title: "#{number_with_delimiter(row.processed)} of #{number_with_delimiter(totals.processed)} processed") %>
+                  <span class="w-10 shrink-0 text-right font-mono text-xs tabular-nums text-zinc-500"><%= share.round %>%</span>
+                </div>
+              </td>
+              <td class="px-4 py-2.5 text-right font-mono tabular-nums">
+                <% if row.in_flight.zero? %>
+                  <span class="text-zinc-300">0</span>
+                <% else %>
+                  <%= link_to number_with_delimiter(row.in_flight), workflows_path(job_class: row.job_class, state: "in_flight", hide_branches: "0"), class: "text-zinc-600 hover:underline" %>
+                <% end %>
+              </td>
+              <td class="px-4 py-2.5 text-right font-mono tabular-nums">
+                <% if row.blocked.zero? %>
+                  <span class="text-zinc-300">0</span>
+                <% else %>
+                  <%= link_to "#{number_with_delimiter(row.blocked)} ⚠", workflows_path(job_class: row.job_class, state: "blocked", hide_branches: "0"), class: "whitespace-nowrap font-medium text-rose-600 hover:underline" %>
+                <% end %>
+              </td>
+            </tr>
+          <% end %>
+        </tbody>
+        <% if @overview.rows.any? %>
+          <tfoot>
+            <tr class="border-t border-zinc-200 bg-zinc-50 font-medium">
+              <td class="px-4 py-2.5">Totals</td>
+              <td class="px-4 py-2.5 text-right font-mono tabular-nums text-zinc-900"><%= number_with_delimiter(totals.processed) %></td>
+              <td class="px-4 py-2.5 text-xs text-zinc-400"><%= @overview.rows.size %> <%= "class".pluralize(@overview.rows.size) %></td>
+              <td class="px-4 py-2.5 text-right font-mono tabular-nums text-zinc-600"><%= number_with_delimiter(totals.in_flight) %></td>
+              <td class="px-4 py-2.5 text-right font-mono tabular-nums <%= totals.blocked > 0 ? "text-rose-600" : "text-zinc-600" %>"><%= number_with_delimiter(totals.blocked) %></td>
+            </tr>
+          </tfoot>
+        <% end %>
+      </table>
+    </div>
+    <% if @overview.rows.none? %>
+      <p class="px-4 py-12 text-center text-sm text-zinc-500">No workflows yet — nothing has been processed.</p>
+    <% end %>
+  </div>
+</turbo-frame>

data/app/views/chrono_forge/dashboard/overview/in_flight.html.erb

@@ -0,0 +1,4 @@
+<turbo-frame id="cf-ov-in-flight">
+  <%= render "stat_card", href: workflows_path(state: "in_flight", hide_branches: "0"),
+        label: "In flight", count: @count, sub: "idle + running", color: "text-zinc-900" %>
+</turbo-frame>

data/app/views/chrono_forge/dashboard/overview/index.html.erb

@@ -0,0 +1,36 @@
+<div class="mb-5">
+  <h1 class="text-lg font-semibold tracking-tight">Overview</h1>
+  <p class="mt-1 text-sm text-zinc-500">
+    How much every workflow class has processed, and what it's carrying right now. Counts are workflows across all time.
+  </p>
+</div>
+
+<%# Fleet totals — each card loads in its own frame (one COUNT), so they appear as
+    soon as they're ready without waiting on the per-class table below. %>
+<%# target="_top": each frame's content is links (a whole card, class names, counts)
+    that must navigate the full page — without this they'd resolve inside the frame
+    and Turbo would report "content missing". It doesn't affect the src load. %>
+<div class="mb-5 grid grid-cols-3 gap-3">
+  <turbo-frame id="cf-ov-processed" src="<%= overview_processed_path %>" loading="eager" target="_top">
+    <%= render "card_skeleton", label: "Processed" %>
+  </turbo-frame>
+  <turbo-frame id="cf-ov-in-flight" src="<%= overview_in_flight_path %>" loading="eager" target="_top">
+    <%= render "card_skeleton", label: "In flight" %>
+  </turbo-frame>
+  <turbo-frame id="cf-ov-blocked" src="<%= overview_blocked_path %>" loading="eager" target="_top">
+    <%= render "card_skeleton", label: "Blocked" %>
+  </turbo-frame>
+</div>
+
+<%# Per-class breakdown — the heavy GROUP BY, isolated in its own frame so it
+    streams in on its own clock. %>
+<turbo-frame id="cf-ov-classes" src="<%= overview_classes_path %>" loading="eager" target="_top">
+  <div class="cf-card p-6">
+    <div class="h-4 w-40 animate-pulse rounded bg-zinc-100"></div>
+    <div class="mt-4 space-y-3">
+      <% 5.times do %>
+        <div class="h-3 w-full animate-pulse rounded bg-zinc-100"></div>
+      <% end %>
+    </div>
+  </div>
+</turbo-frame>

data/app/views/chrono_forge/dashboard/overview/processed.html.erb

@@ -0,0 +1,4 @@
+<turbo-frame id="cf-ov-processed">
+  <%= render "stat_card", href: workflows_path(state: "completed", hide_branches: "0"),
+        label: "Processed", count: @count, sub: "completed, all classes", color: "text-emerald-600" %>
+</turbo-frame>

data/app/views/chrono_forge/dashboard/repetitions/index.html.erb

@@ -28,29 +28,51 @@
         <th class="px-4 py-2.5 font-medium">Error</th>
       </tr>
     </thead>
+    <%# One scale per page so the meters below are comparable: the latest run's
+        biggest lateness and longest duration. In-memory over the loaded page. %>
+    <% ts_for = ->(run) { ChronoForge::Dashboard::StepNameParser.parse(run.step_name).timestamp } %>
+    <% max_late = @runs.filter_map { |r| t = ts_for.call(r); (r.started_at - Time.zone.at(t)).to_i if t && r.started_at }.select(&:positive?).max || 1 %>
+    <% max_dur = @runs.filter_map { |r| (r.completed_at - r.started_at).to_i if r.started_at && r.completed_at }.max || 1 %>
     <tbody class="divide-y divide-zinc-100">
       <% @runs.each do |run| %>
-        <% ts = ChronoForge::Dashboard::StepNameParser.parse(run.step_name).timestamp %>
+        <% ts = ts_for.call(run) %>
         <% ff = run.failed? ? run.metadata&.dig("fast_forwarded") : nil %>
         <% tombstone = run.failed? && ff.nil? %>
         <tr class="<%= "bg-amber-50" if run.failed? %>">
           <td class="px-4 py-2.5 font-mono text-xs text-zinc-600"><%= ts ? cf_ago(Time.zone.at(ts)) : "—" %></td>
           <td class="px-4 py-2.5 text-xs">
             <% if ff %>
-              <span class="font-medium text-amber-700" title="fast-forwarded <%= ff %> expired tick(s): <%= run.metadata["from"] %> → <%= run.metadata["to"] %>">caught up ×<%= ff %></span>
+              <span class="cf-pill cf-pill-stalled" title="fast-forwarded <%= ff %> expired tick(s): <%= run.metadata["from"] %> → <%= run.metadata["to"] %>">caught up ×<%= ff %></span>
             <% elsif tombstone %>
-              <span class="font-medium text-amber-700">tombstone</span>
+              <span class="cf-pill cf-pill-stalled" title="an expired tick the engine stepped past — normal catch-up, not an error">tombstone</span>
+            <% elsif run.completed? %>
+              <span class="cf-pill cf-pill-completed">done</span>
             <% else %>
-              <span class="<%= cf_status_color(run.state) %>"><%= run.state %></span>
+              <span class="cf-pill cf-pill-running"><%= run.state %></span>
             <% end %>
           </td>
           <td class="px-4 py-2.5 text-zinc-600"><%= cf_ago(run.started_at) %></td>
           <% late = (ts && run.started_at) ? (run.started_at - Time.zone.at(ts)).to_i : nil %>
-          <td class="px-4 py-2.5 text-right font-mono text-xs <%= (late && late > 60) ? "text-amber-600" : "text-zinc-500" %>">
-            <%= late.nil? ? "—" : (late <= 0 ? "on time" : cf_secs(late)) %>
+          <% late_amber = late && late > 60 %>
+          <td class="px-4 py-2.5">
+            <div class="flex items-center justify-end gap-2">
+              <% if late&.positive? %><%= cf_meter(cf_bar_width(late, max_late), late_amber ? "bg-amber-400" : "bg-zinc-400", track: "w-11", title: "started #{cf_secs(late)} after its scheduled tick") %><% end %>
+              <span class="whitespace-nowrap text-right font-mono text-xs <%= late_amber ? "font-medium text-amber-600" : "text-zinc-500" %>"><%= late.nil? ? "—" : (late <= 0 ? "on time" : cf_secs(late)) %></span>
+            </div>
+          </td>
+          <% dur = (run.started_at && run.completed_at) ? (run.completed_at - run.started_at).to_i : nil %>
+          <td class="px-4 py-2.5">
+            <div class="flex items-center justify-end gap-2">
+              <% if dur %><%= cf_meter(cf_duration_bar(dur, max_dur), cf_slow_step?(dur) ? "bg-amber-400" : "bg-zinc-400", track: "w-11", title: cf_meter_title(dur, max_dur, run.started_at, run.completed_at)) %><% end %>
+              <span class="whitespace-nowrap text-right font-mono text-xs <%= cf_slow_step?(dur) ? "font-medium text-amber-600" : "text-zinc-500" %>"><%= cf_duration(run.started_at, run.completed_at) %></span>
+            </div>
+          </td>
+          <td class="px-4 py-2.5 text-right">
+            <% if (note = cf_attempts_note(:execute, run.attempts, run.state)) %>
+              <% tone = {muted: "border-zinc-200 bg-zinc-50 text-zinc-500", warn: "border-amber-200 bg-amber-50 text-amber-700", crit: "border-rose-200 bg-rose-50 text-rose-700"}[note[:tone]] %>
+              <span class="rounded border px-1.5 py-0.5 font-mono text-[11px] <%= tone %>" title="<%= note[:title] %>"><%= note[:text] %></span>
+            <% end %>
           </td>
-          <td class="px-4 py-2.5 text-right font-mono text-xs text-zinc-500"><%= cf_duration(run.started_at, run.completed_at) %></td>
-          <td class="px-4 py-2.5 text-right font-mono text-xs text-zinc-500">×<%= run.attempts %></td>
           <td class="px-4 py-2.5 font-mono text-xs text-rose-600"><%= ff ? "—" : (run.error_class || "—") %></td>
         </tr>
       <% end %>

data/app/views/chrono_forge/dashboard/stranded/index.html.erb

@@ -0,0 +1,73 @@
+<div class="mb-1 flex flex-wrap items-center justify-between gap-3">
+  <h1 class="text-lg font-semibold tracking-tight">Stranded workflows</h1>
+  <% if @stranded.any? %>
+    <%# Background sweep — mirrors "Retry blocked". Disables + relabels while the
+        POST runs and survives the poll morph (see dashboard.js). %>
+    <%= button_to "Reap all stranded", reap_all_stranded_index_path, method: :post,
+          form: {data: {confirm: "Reap every stranded workflow? Each is re-enqueued to steal its stale lock and replay; steps with external side effects must be idempotent.", cf_poll_preserve: true}},
+          data: {turbo_submits_with: "Reaping…"}, class: "cf-btn cf-btn-primary" %>
+  <% end %>
+</div>
+<p class="mb-5 text-sm text-zinc-500">
+  Workflows still in <span class="font-mono">running</span> whose lock hasn't been refreshed in
+  <%= cf_secs(@stale_after.to_i) %> (<span class="font-mono">reap_stale_after</span>) — no worker is driving them.
+  A worker hard-killed mid-pass (SIGKILL / OOM / eviction) leaves the row locked with nothing scheduled to wake it.
+  <strong class="font-medium text-zinc-600">Reaping</strong> re-enqueues each so the executor steals the stale lock and replays completed steps as no-ops.
+</p>
+
+<% if @stranded.any? %>
+  <div class="mb-5 flex items-start gap-3 rounded-lg border border-rose-200 bg-rose-50 px-4 py-3">
+    <span class="mt-px text-rose-600" aria-hidden="true">▲</span>
+    <div class="min-w-0 flex-1 text-sm">
+      <p class="font-semibold text-zinc-900"><%= @capped ? "#{@cap}+" : @stranded.size %> stranded <%= "workflow".pluralize(@stranded.size) %> to reap</p>
+      <p class="mt-0.5 text-zinc-600">These match <span class="font-mono">Workflow.reap_stalled</span> exactly — the same sweep a periodic reaper would run. Reap them here, or let the scheduled sweep pick them up.</p>
+    </div>
+  </div>
+<% end %>
+
+<div class="cf-card overflow-hidden">
+  <div class="overflow-x-auto">
+    <table class="w-full min-w-[44rem] text-sm">
+      <thead>
+        <tr class="border-b border-zinc-200 bg-zinc-50 text-left text-xs uppercase tracking-wide text-zinc-500">
+          <th class="px-4 py-2.5 font-medium">Key</th>
+          <th class="px-4 py-2.5 font-medium">Class</th>
+          <th class="px-4 py-2.5 font-medium">Locked by</th>
+          <th class="px-4 py-2.5 font-medium">Lock stale for</th>
+          <th class="px-4 py-2.5 text-right font-medium">Started</th>
+          <th class="px-4 py-2.5 text-right font-medium">Reap</th>
+        </tr>
+      </thead>
+      <tbody class="divide-y divide-zinc-100">
+        <% max_age = @stranded.filter_map { |wf| cf_lock_age(wf) }.max || 1 %>
+        <% @stranded.each do |wf| %>
+          <% age = cf_lock_age(wf) %>
+          <tr class="bg-rose-50/40 hover:bg-rose-50">
+            <td class="px-4 py-2.5"><%= link_to wf.key, workflow_path(wf), class: "font-mono hover:underline" %></td>
+            <td class="px-4 py-2.5 font-mono text-xs text-zinc-600"><%= wf.job_class %></td>
+            <td class="px-4 py-2.5 font-mono text-xs text-zinc-500"><%= wf.locked_by || "—" %></td>
+            <td class="px-4 py-2.5">
+              <div class="flex items-center gap-2">
+                <%= cf_meter(cf_bar_width(age, max_age), "bg-rose-400", track: "w-14", title: "lock last refreshed #{cf_secs(age)} ago") %>
+                <span class="whitespace-nowrap font-medium text-rose-700"><%= cf_duration(wf.locked_at, Time.current) %></span>
+              </div>
+            </td>
+            <td class="px-4 py-2.5 text-right text-xs text-zinc-500"><%= cf_ago(wf.started_at) %></td>
+            <td class="px-4 py-2.5 text-right">
+              <%= button_to "Reap", reap_workflow_path(wf), method: :post,
+                    form: {data: {confirm: "Reap #{wf.key}? It re-enqueues the run to steal its stale lock and replay. Steps with external side effects must be idempotent.", cf_poll_preserve: true}},
+                    data: {turbo_submits_with: "Reaping…"}, class: "cf-btn" %>
+            </td>
+          </tr>
+        <% end %>
+      </tbody>
+    </table>
+  </div>
+  <% if @stranded.none? %>
+    <p class="px-4 py-12 text-center text-sm text-zinc-500">No stranded workflows — every running workflow has a live lock. 🎉</p>
+  <% end %>
+</div>
+
+<% if @capped %>
+  <p class="mt-3 text-xs text-zinc-400">Showing the <%= @cap %> longest-stranded. "Reap all stranded" sweeps every one, not just these.</p>
+<% end %>

data/app/views/chrono_forge/dashboard/wait_states/index.html.erb

@@ -5,6 +5,20 @@
   arrives sits here indefinitely.
 </p>
 
+<%# Lead with the problem: event waits that have blown past the threshold. These
+    never time out, so nothing else will surface them. %>
+<% stalled_events = @waits.select { |h| h[:wait].event_wait? && h[:wait].waiting_since && (Time.current - h[:wait].waiting_since) > @threshold } %>
+<% if stalled_events.any? %>
+  <% oldest = stalled_events.first %>
+  <div class="mb-5 flex items-start gap-3 rounded-lg border border-rose-200 bg-rose-50 px-4 py-3">
+    <span class="mt-px text-rose-600" aria-hidden="true">▲</span>
+    <div class="min-w-0 flex-1 text-sm">
+      <p class="font-semibold text-zinc-900"><%= pluralize(stalled_events.size, "event-wait") %> stalled beyond <%= cf_secs(@threshold) %></p>
+      <p class="mt-0.5 text-zinc-600">A <span class="font-mono">continue_if</span> waits on an external signal with no timeout — if it never fires, the workflow waits forever. Oldest: <span class="font-mono text-rose-600"><%= oldest[:workflow].job_class %> · <%= oldest[:wait].condition %></span>, waiting <%= distance_of_time_in_words(oldest[:wait].waiting_since, Time.current) %>.</p>
+    </div>
+  </div>
+<% end %>
+
 <% if @oldest_event_by_class.any? %>
   <section class="cf-card mb-5 border-amber-200 bg-amber-50/40 p-5">
     <h2 class="mb-1 text-xs font-medium uppercase tracking-wide text-amber-700">Oldest unresolved event wait, by class</h2>
@@ -44,20 +58,30 @@
     </thead>
     <tbody class="divide-y divide-zinc-100">
       <% @waits.each do |h| %>
-        <% long = (Time.current - (h[:wait].waiting_since || Time.current)) > @threshold %>
-        <tr class="cursor-pointer <%= "cf-wait--long bg-amber-50" if long %> hover:bg-zinc-50" data-href="<%= workflow_path(h[:workflow]) %>">
+        <% wait = h[:wait] %>
+        <% age = wait.waiting_since ? (Time.current - wait.waiting_since).to_i : nil %>
+        <% long = age && age > @threshold %>
+        <%# An over-threshold event wait is the urgent case (no timeout); a long
+            poll is merely slow. Escalate the former to rose, the latter amber. %>
+        <% urgent = long && wait.event_wait? %>
+        <tr class="cursor-pointer <%= urgent ? "cf-wait--long bg-rose-50" : "cf-wait--long bg-amber-50" if long %> hover:bg-zinc-50" data-href="<%= workflow_path(h[:workflow]) %>">
           <td class="px-4 py-2.5"><%= link_to h[:workflow].key, workflow_path(h[:workflow]), class: "font-mono hover:underline" %></td>
           <td class="px-4 py-2.5 font-mono text-xs text-zinc-600"><%= h[:workflow].job_class %></td>
           <td class="px-4 py-2.5 text-xs">
-            <% if h[:wait].event_wait? %>
-              <span class="font-medium text-amber-700">event</span>
+            <% if wait.event_wait? %>
+              <span class="cf-pill cf-pill-stalled" title="continue_if — waits on an external event, no timeout">event</span>
             <% else %>
-              <span class="text-zinc-500">poll</span>
+              <span class="cf-pill cf-pill-idle" title="wait_until — polls on a schedule, times out">poll</span>
             <% end %>
           </td>
-          <td class="px-4 py-2.5 font-mono text-xs text-zinc-600"><%= h[:wait].condition %></td>
-          <td class="px-4 py-2.5 <%= long ? "font-medium text-amber-700" : "text-zinc-600" %>"><%= distance_of_time_in_words(h[:wait].waiting_since, Time.current) %></td>
-          <td class="px-4 py-2.5 text-right font-mono text-xs text-zinc-500"><%= h[:wait].event_wait? ? "none" : (h[:wait].timeout_at || "—") %></td>
+          <td class="px-4 py-2.5 font-mono text-xs text-zinc-600"><%= wait.condition %></td>
+          <td class="px-4 py-2.5">
+            <div class="flex items-center gap-2">
+              <% if age %><%= cf_meter(cf_bar_width([age, @threshold].min, @threshold), urgent ? "bg-rose-400" : long ? "bg-amber-400" : "bg-zinc-400", track: "w-12", title: "waiting #{cf_secs(age)} of the #{cf_secs(@threshold)} threshold") %><% end %>
+              <span class="whitespace-nowrap text-sm <%= urgent ? "font-medium text-rose-700" : long ? "font-medium text-amber-700" : "text-zinc-600" %>"><%= distance_of_time_in_words(wait.waiting_since, Time.current) %></span>
+            </div>
+          </td>
+          <td class="px-4 py-2.5 text-right font-mono text-xs text-zinc-500"><%= wait.event_wait? ? "none" : (wait.timeout_at || "—") %></td>
         </tr>
       <% end %>
     </tbody>

data/app/views/chrono_forge/dashboard/workflows/_error_card.html.erb

@@ -7,7 +7,7 @@
   <% if err.backtrace.present? %>
     <details class="mt-1">
       <summary class="cursor-pointer text-[11px] text-zinc-500">backtrace</summary>
-      <pre class="mt-1 overflow-x-auto rounded bg-white p-2 font-mono text-[11px] text-zinc-600"><%= err.backtrace %></pre>
+      <pre class="mt-1 max-h-64 overflow-auto rounded bg-white p-2 font-mono text-[11px] text-zinc-600"><%= err.backtrace %></pre>
     </details>
   <% end %>
 </div>

data/app/views/chrono_forge/dashboard/workflows/_filters.html.erb

@@ -1,8 +1,14 @@
 <%# enforce_utf8: false keeps the legacy `utf8=✓` IE hack out of the GET query string. %>
 <%= form_with url: workflows_path, method: :get, enforce_utf8: false, class: "mb-4 flex flex-wrap items-center gap-2" do |f| %>
-  <%= f.select :state, options_for_select([["All states", ""], *ChronoForge::Workflow.states.keys], params[:state]), {}, class: "rounded-md border border-zinc-300 bg-white px-2 py-1.5 text-sm", data: {autosubmit: true} %>
-  <%= f.text_field :job_class, value: params[:job_class], placeholder: "Job class", class: "rounded-md border border-zinc-300 px-2.5 py-1.5 text-sm placeholder:text-zinc-400" %>
-  <%= f.text_field :key, value: params[:key], placeholder: "Key", class: "rounded-md border border-zinc-300 px-2.5 py-1.5 text-sm placeholder:text-zinc-400" %>
+  <%# State filtering lives in the stat chips above (one click, with counts and the
+      blocked virtual filter) — this hidden field just carries the chip-selected
+      state through a job-class/key submit so the search doesn't clear it. %>
+  <%= f.hidden_field :state, value: params[:state] %>
+  <%# data-cf-poll-preserve: the polling morph refresh (see dashboard.js) skips
+      these nodes in place, so text a user is typing — and its caret — survives a
+      tick instead of being reset to the last-submitted server value. %>
+  <%= f.text_field :job_class, value: params[:job_class], placeholder: "Job class", id: "cf-filter-job-class", data: {cf_poll_preserve: true}, class: "rounded-md border border-zinc-300 px-2.5 py-1.5 text-sm placeholder:text-zinc-400" %>
+  <%= f.text_field :key, value: params[:key], placeholder: "Key", id: "cf-filter-key", data: {cf_poll_preserve: true}, class: "rounded-md border border-zinc-300 px-2.5 py-1.5 text-sm placeholder:text-zinc-400" %>
   <%= f.submit "Filter", class: "cf-btn" %>
 
   <%# Far right: hide spawned branch children (on by default) so a large fan-out