roundhouse_ui 0.6.0 → 0.7.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 73ff2f1bc8db03a06168ee39d18b1de518e1a373860e1e0faafdbd08370d5eab
-  data.tar.gz: e1bb6668de35f02e64c369d016a21af989984f604a6009efba39160bd05b5b6a
+  metadata.gz: 5064ee74f5aea444f5762d2f466f0b4c05d9c7af6c44a48045e6703d48b72587
+  data.tar.gz: 3b5a08f1769a5bd5ee66bdbe5ccd586842373b45530b95cbc4ff6340fe17796e
 SHA512:
-  metadata.gz: a27aa6ed4d0a91982e1997a6dcb6492a734edbfcbe3c4cf1c914366b4019b8ba49e922c7bfe59a45da662493d12b196c1f5bcd5ac47f9486248458877f42bb7f
-  data.tar.gz: faa3b64f3da583b3e361b431b8e6fa7b89b4c077a07cc981d54dfd2b20f248144c2798a1497aae4220de86fe385aca7f335b15d6b63c9d9648b22aa05f4acde5
+  metadata.gz: 6d3cf1e161baec06cb4b899d9f97065fb216341716719756c062a140a811bf51997748f9bcc322dc78aaa8ec36d56311decc511b70eea8d53a07a57a5b317b8c
+  data.tar.gz: 15a01ddaa5fa81c37c67516c369d1dac2b795b011e5dad1927b3b70aeed47b365a4938765e78f9eae910da0a39e36fe174d6170ba0226d1eeab404a700f4c218

data/app/controllers/roundhouse_ui/dashboard_controller.rb

@@ -5,6 +5,11 @@ module RoundhouseUi
     def show
       @stats  = Sidekiq::Stats.new
       @queues = Sidekiq::Queue.all
+      @metrics = Metrics.new(stats: @stats)
+      @health  = Health.new(stats: @stats, queues: @queues, metrics: @metrics)
+      # Highest-signal slices for the overview, from data we already read.
+      @top_errors = ErrorGroups.new(limit: 200).call.groups.first(5)
+      @problem_queues = @queues.select { |q| q.latency > 5 }.sort_by { |q| -q.latency }.first(5)
     end
 
     # Polled by the dashboard for live counts (same approach Sidekiq Web uses —

data/app/controllers/roundhouse_ui/errors_controller.rb

@@ -3,59 +3,11 @@ module RoundhouseUi
   # (job class + error class) — so one bad deploy reads as a single issue with
   # a count, not five thousand identical rows. The aggregation Sidekiq Web lacks.
   class ErrorsController < ApplicationController
-    SCAN_LIMIT = 1_000 # cap entries scanned per pass; shown honestly in the view
-
     def index
       @query = params[:q].to_s.strip
-      @scan_limit = SCAN_LIMIT
-      @groups, @scanned, @truncated = aggregate
-    end
-
-    private
-
-    def aggregate
-      groups = {}
-      scanned = 0
-      truncated = false
-
-      sources.each do |source, set|
-        set.each do |entry|
-          scanned += 1
-          if scanned > SCAN_LIMIT
-            truncated = true
-            break
-          end
-          record(groups, source, entry)
-        end
-        break if truncated
-      end
-
-      list = groups.values.sort_by { |g| -g[:count] }
-      list = list.select { |g| "#{g[:klass]} #{g[:error]}".downcase.include?(@query.downcase) } if @query.present?
-      [ list, scanned, truncated ]
-    end
-
-    # Sidekiq's native sets, plus the sidekiq-failures `failed` set when opted in
-    # and that gem is loaded. Its FailureSet is a Sidekiq::JobSet, so it iterates
-    # exactly like the others — no special-casing in the aggregation above.
-    def sources
-      sets = { "retry" => Sidekiq::RetrySet.new, "dead" => Sidekiq::DeadSet.new }
-      if RoundhouseUi.show_sidekiq_failures && defined?(Sidekiq::Failures::FailureSet)
-        sets["failed"] = Sidekiq::Failures::FailureSet.new
-      end
-      sets
-    end
-
-    def record(groups, source, entry)
-      error = entry.item["error_class"] || "UnknownError"
-      group = (groups["#{entry.klass}|#{error}"] ||= {
-        klass: entry.klass, error: error, count: 0, last_at: nil, queues: [], sources: []
-      })
-      group[:count]  += 1
-      group[:queues] |= [ entry.queue ]
-      group[:sources] |= [ source ]
-      at = entry.at
-      group[:last_at] = at if at && (group[:last_at].nil? || at > group[:last_at])
+      @scan_limit = ErrorGroups::DEFAULT_SCAN_LIMIT
+      result = ErrorGroups.new(query: @query).call
+      @groups, @scanned, @truncated = result.groups, result.scanned, result.truncated
     end
   end
 end

data/app/views/layouts/roundhouse_ui/application.html.erb

@@ -117,12 +117,52 @@
     .rh-card .d.up { color:var(--good); } .rh-card .d.bad { color:var(--crit); }
     .rh-card .pill { display:inline-flex; align-items:center; gap:6px; font-size:13px; font-weight:600; padding:4px 10px; border-radius:7px; margin-bottom:9px; background:rgba(226,165,63,.14); color:var(--warn); }
     .rh-card .pill.ok { background:rgba(68,197,140,.14); color:var(--good); }
-    .rh-chart-wrap { background:var(--panel); border:1px solid var(--line-soft); border-radius:12px; padding:18px 20px 10px; margin-bottom:24px; }
+    .rh-chart-wrap { background:var(--panel); border:1px solid var(--line-soft); border-radius:12px; padding:18px 20px 10px; margin-bottom:24px; position:relative; }
+    #rh-chart-tip { position:absolute; top:34px; transform:translateX(-50%); pointer-events:none; background:var(--panel-3); border:1px solid var(--line); color:var(--text); font:11px var(--mono); padding:2px 7px; border-radius:6px; white-space:nowrap; }
+    #rh-chart-tip[hidden] { display:none; }
     .rh-chart-wrap .top { display:flex; align-items:baseline; gap:12px; margin-bottom:6px; }
     .rh-chart-wrap h3 { font-size:13px; font-weight:600; margin:0; }
     .rh-chart-wrap .now { margin-left:auto; font-family:var(--mono); font-size:13px; color:var(--accent); }
     canvas#rh-chart { display:block; width:100%; height:90px; }
     .rh-cb { width:15px; height:15px; }
+
+    /* composite health banner */
+    .rh-health { background:var(--panel); border:1px solid var(--line-soft); border-left:3px solid var(--faint); border-radius:12px; margin-bottom:18px; }
+    .rh-health-ok { border-left-color:var(--good); }
+    .rh-health-warn { border-left-color:var(--warn); }
+    .rh-health-crit { border-left-color:var(--crit); }
+    .rh-health summary { display:flex; align-items:center; gap:12px; padding:14px 16px; cursor:pointer; list-style:none; }
+    .rh-health summary::-webkit-details-marker { display:none; }
+    .rh-health-dot { width:9px; height:9px; border-radius:50%; flex:none; background:var(--faint); }
+    .rh-health-ok .rh-health-dot { background:var(--good); }
+    .rh-health-warn .rh-health-dot { background:var(--warn); }
+    .rh-health-crit .rh-health-dot { background:var(--crit); }
+    .rh-health-verdict { font-weight:650; font-size:15px; }
+    .rh-health-ok .rh-health-verdict { color:var(--good); }
+    .rh-health-warn .rh-health-verdict { color:var(--warn); }
+    .rh-health-crit .rh-health-verdict { color:var(--crit); }
+    .rh-health-reason { color:var(--muted); font-size:13px; }
+    .rh-health-cta { margin-left:auto; font:11px var(--mono); color:var(--faint); }
+    .rh-health[open] .rh-health-cta { color:var(--accent); }
+    .rh-health-signals { padding:2px 16px 14px; display:flex; flex-direction:column; gap:8px; border-top:1px solid var(--line-soft); margin-top:2px; padding-top:12px; }
+    .rh-sig { display:flex; align-items:center; gap:10px; font-size:13px; }
+    .rh-sig b { min-width:170px; font-weight:600; }
+
+    /* insight panels */
+    .rh-insights { display:grid; gap:14px; margin-bottom:24px; }
+    @media(min-width:760px){ .rh-insights { grid-template-columns:1fr 1fr; } }
+    .rh-insight { padding:6px 0; }
+    .rh-insight-h { display:flex; align-items:baseline; justify-content:space-between; gap:10px; font-size:13px; font-weight:600; padding:13px 16px 10px; }
+    .rh-insight-row { display:flex; align-items:center; gap:12px; padding:10px 16px; border-top:1px solid var(--line-soft); }
+    .rh-insight-row .rh-insight-main { flex:1; min-width:0; font-size:13px; }
+    .rh-insight-row .rh-pill { font-variant-numeric:tabular-nums; }
+    .rh-insight .rh-empty { padding:18px 16px; }
+
+    /* collapsible disclosure (large args, full backtrace) */
+    .rh-disclose summary { cursor:pointer; font:12px var(--mono); color:var(--accent); padding:6px 0; list-style:none; }
+    .rh-disclose summary::-webkit-details-marker { display:none; }
+    .rh-disclose summary::before { content:"▸ "; }
+    .rh-disclose[open] summary::before { content:"▾ "; }
     .rh-field { margin-bottom:16px; max-width:640px; }
     .rh-field label { display:block; font-size:12px; color:var(--muted); margin-bottom:6px; }
     .rh-field input, .rh-field textarea { width:100%; background:var(--panel); border:1px solid var(--line); border-radius:9px; padding:9px 12px; color:var(--text); font:13px var(--mono); }
@@ -160,6 +200,7 @@
       var started = false, lastProcessed = null, lastFailed = null, lastBacklog = null, lastT = null;
       var POLL_MS = <%= (RoundhouseUi.poll_interval.to_f * 1000).round %>;
       var samples = [], buckets = [], bucketStart = null; // samples = current bucket; buckets = finalized per-interval averages
+      var lastPts = []; // points last drawn, for the hover tooltip to read
       function setText(id, t) { var el = document.getElementById(id); if (el) el.textContent = t; }
       function humanizeEta(s) {
         if (s < 60) return "~" + Math.round(s) + "s";
@@ -179,12 +220,28 @@
         // finalized buckets + the in-progress bucket as a provisional last point,
         // so the chart shows data immediately instead of waiting a full interval.
         var pts = buckets.slice(-60); if (samples.length) pts = pts.concat([ avg(samples) ]);
-        var n = pts.length; if (n < 2) return;
-        var max = Math.max.apply(null, pts) * 1.25 || 1;
+        lastPts = pts;
+        var n = pts.length; if (n < 2) { setText("rh-chart-peak", ""); return; }
+        var peak = Math.max.apply(null, pts);
+        var max = peak * 1.25 || 1;
         var x = function (i) { return i / (n - 1) * w; }, y = function (v) { return h - pad - v / max * (h - pad * 2); };
         var g = ctx.createLinearGradient(0, 0, 0, h); g.addColorStop(0, "rgba(110,139,255,.30)"); g.addColorStop(1, "rgba(110,139,255,0)");
         ctx.beginPath(); ctx.moveTo(0, h); pts.forEach(function (v, i) { ctx.lineTo(x(i), y(v)); }); ctx.lineTo(w, h); ctx.closePath(); ctx.fillStyle = g; ctx.fill();
         ctx.beginPath(); pts.forEach(function (v, i) { i ? ctx.lineTo(x(i), y(v)) : ctx.moveTo(x(i), y(v)); }); ctx.strokeStyle = "#6E8BFF"; ctx.lineWidth = 2; ctx.lineJoin = "round"; ctx.stroke();
+        // emphasized endpoint (small marker; canvas is vertically scaled, so a rect reads cleaner than an arc)
+        ctx.fillStyle = "#6E8BFF"; ctx.fillRect(x(n - 1) - 3, y(pts[n - 1]) - 3, 6, 6);
+        setText("rh-chart-peak", "peak " + Math.round(peak) + "/s");
+      }
+      // Hover tooltip: map the cursor's fractional x to the nearest drawn point.
+      function chartHover(e) {
+        var cv = document.getElementById("rh-chart"), tip = document.getElementById("rh-chart-tip");
+        if (!cv || !tip || lastPts.length < 2) return;
+        var rect = cv.getBoundingClientRect();
+        var fx = Math.min(1, Math.max(0, (e.clientX - rect.left) / rect.width));
+        var idx = Math.round(fx * (lastPts.length - 1));
+        tip.textContent = Math.round(lastPts[idx]) + "/s";
+        tip.style.left = (e.clientX - rect.left) + "px";
+        tip.hidden = false;
       }
       function apply(d) {
         Object.keys(d).forEach(function (k) {
@@ -231,6 +288,11 @@
           .catch(function () {});
       }
       function startOnce() { if (started) return; started = true; poll(); setInterval(poll, POLL_MS); }
+      function wireChart() {
+        var cv = document.getElementById("rh-chart"); if (!cv) return;
+        cv.onmousemove = chartHover;
+        cv.onmouseleave = function () { var t = document.getElementById("rh-chart-tip"); if (t) t.hidden = true; };
+      }
       function syncTheme() { var b = document.getElementById("rh-theme"); if (b) b.textContent = document.documentElement.getAttribute("data-theme") === "light" ? "☀" : "☾"; }
       function syncWidth() { var b = document.getElementById("rh-width"); if (b) b.classList.toggle("is-on", document.documentElement.getAttribute("data-width") === "full"); }
 
@@ -342,8 +404,8 @@
         else if (e.key === "Enter") { e.preventDefault(); if (palFiltered[palSel]) palRun(palFiltered[palSel]); }
       });
 
-      document.addEventListener("turbo:load", function () { startOnce(); syncTheme(); syncWidth(); setActiveNav(); restoreChartInterval(); draw(); });
-      document.addEventListener("DOMContentLoaded", function () { startOnce(); syncTheme(); syncWidth(); setActiveNav(); restoreChartInterval(); });
+      document.addEventListener("turbo:load", function () { startOnce(); syncTheme(); syncWidth(); setActiveNav(); restoreChartInterval(); wireChart(); draw(); });
+      document.addEventListener("DOMContentLoaded", function () { startOnce(); syncTheme(); syncWidth(); setActiveNav(); restoreChartInterval(); wireChart(); });
       document.addEventListener("visibilitychange", function () { if (!document.hidden) poll(); });
     })();
   </script>

data/app/views/roundhouse_ui/dashboard/show.html.erb

@@ -1,49 +1,83 @@
 <% content_for :title, "Dashboard" %>
 <% content_for :crumb, Rails.env %>
 
-<% stuck = @queues.select { |q| q.latency > 60 } %>
-
-<% if stuck.any? %>
-  <div class="rh-alerts">
-    <% stuck.each do |q| %>
-      <div class="rh-alert <%= "warn" if q.latency <= 600 %>">
-        <span class="msg">Queue <b><%= q.name %></b> is <%= q.latency > 600 ? "stuck" : "over budget" %> — oldest job <%= distance_of_time_in_words(0, q.latency) %>, <%= number_with_delimiter q.size %> waiting</span>
-        <%= link_to "Manage →", queues_path, class: "rh-btn" %>
+<% verdict = { ok: "Healthy", warn: "Degraded", crit: "Critical" }[@health.status] %>
+<details class="rh-health rh-health-<%= @health.status %>"<%= " open".html_safe unless @health.healthy? %>>
+  <summary>
+    <span class="rh-health-dot"></span>
+    <span class="rh-health-verdict"><%= verdict %></span>
+    <span class="rh-health-reason"><%= @health.reason %></span>
+    <span class="rh-health-cta">why ▾</span>
+  </summary>
+  <div class="rh-health-signals">
+    <% @health.signals.each do |s| %>
+      <div class="rh-sig">
+        <span class="rh-st rh-st-<%= s.status %>"><%= s.status %></span>
+        <b><%= s.label %></b>
+        <span class="rh-sub"><%= s.detail %></span>
       </div>
     <% end %>
   </div>
-<% end %>
+</details>
 
 <div class="rh-cards">
-  <div class="rh-card">
-    <% if stuck.any? %>
-      <span class="pill">⚠ Degraded</span>
-      <div class="k"><b style="color:var(--warn)"><%= stuck.first.name %></b> queue over budget</div>
-    <% else %>
-      <span class="pill ok">✓ Healthy</span>
-      <div class="k">all queues within budget</div>
-    <% end %>
-  </div>
   <div class="rh-card">
     <div class="k">Processed</div>
     <div class="v num" data-stat="processed"><%= number_with_delimiter @stats.processed %></div>
-    <div class="d"><span class="num" data-stat="rate">—</span> / min</div>
+    <div class="d"><span class="num" data-stat="rate">—</span> / min · <%= ((1 - @metrics.failure_ratio) * 100).round(1) %>% ok</div>
   </div>
   <div class="rh-card">
     <div class="k">Failed · total</div>
     <div class="v num" data-stat="failed"><%= number_with_delimiter @stats.failed %></div>
-    <div class="d bad"><span class="num" data-stat="dead"><%= number_with_delimiter @stats.dead_size %></span> dead</div>
+    <div class="d bad"><span class="num" data-stat="dead"><%= number_with_delimiter @stats.dead_size %></span> dead · <%= (@metrics.failure_ratio * 100).round(1) %>% rate</div>
   </div>
   <div class="rh-card">
     <div class="k">Busy threads</div>
     <div class="v num" data-stat="busy"><%= @stats.workers_size %></div>
-    <div class="d">enqueued <span class="num" data-stat="enqueued"><%= number_with_delimiter @stats.enqueued %></span></div>
+    <div class="d"><%= @metrics.utilization ? "#{(@metrics.utilization * 100).round}% of #{@metrics.concurrency} busy" : "no workers reporting" %></div>
+  </div>
+  <div class="rh-card">
+    <div class="k">Backlog</div>
+    <div class="v num"><%= number_with_delimiter @metrics.backlog %></div>
+    <div class="d">enqueued <span class="num" data-stat="enqueued"><%= number_with_delimiter @stats.enqueued %></span> · scheduled <span class="num" data-stat="scheduled"><%= number_with_delimiter @stats.scheduled_size %></span></div>
   </div>
 </div>
 
 <div class="rh-chart-wrap">
-  <div class="top"><h3>Throughput</h3><span class="rh-sub">jobs / sec · live</span><span class="now"><span id="rh-chart-now">—</span>/s</span><select id="rh-chart-interval" aria-label="Throughput interval" style="margin-left:12px;background:var(--panel-2);color:var(--muted);border:1px solid var(--line);border-radius:7px;padding:3px 6px;font:12px var(--sans)"><option value="10">per 10s</option><option value="30" selected>per 30s</option><option value="60">per 1m</option><option value="300">per 5m</option></select></div>
+  <div class="top"><h3>Throughput</h3><span class="rh-sub">jobs / sec · live</span><span class="rh-sub" id="rh-chart-peak" style="margin-left:10px"></span><span class="now"><span id="rh-chart-now">—</span>/s</span><select id="rh-chart-interval" aria-label="Throughput interval" style="margin-left:12px;background:var(--panel-2);color:var(--muted);border:1px solid var(--line);border-radius:7px;padding:3px 6px;font:12px var(--sans)"><option value="10">per 10s</option><option value="30" selected>per 30s</option><option value="60">per 1m</option><option value="300">per 5m</option></select></div>
   <canvas id="rh-chart" width="1100" height="180"></canvas>
+  <div id="rh-chart-tip" hidden></div>
+</div>
+
+<div class="rh-insights">
+  <div class="rh-panel rh-insight">
+    <div class="rh-insight-h">Top failing job classes <%= link_to "all errors →", errors_path, class: "rh-sub" %></div>
+    <% if @top_errors.empty? %>
+      <div class="rh-empty">No failing jobs 🎉</div>
+    <% else %>
+      <% @top_errors.each do |g| %>
+        <div class="rh-insight-row">
+          <div class="rh-insight-main"><%= link_to g[:klass], errors_path(q: g[:klass]), class: "rh-joblink" %> <%= error_trace_link(klass: g[:klass], error: g[:error]) %><br><span class="rh-err"><%= g[:error] %></span></div>
+          <span class="rh-pill"><%= number_with_delimiter g[:count] %></span>
+        </div>
+      <% end %>
+    <% end %>
+  </div>
+
+  <div class="rh-panel rh-insight">
+    <div class="rh-insight-h">Problem queues <%= link_to "all queues →", queues_path, class: "rh-sub" %></div>
+    <% if @problem_queues.empty? %>
+      <div class="rh-empty">All queues fresh — nothing backing up.</div>
+    <% else %>
+      <% @problem_queues.each do |q| %>
+        <% label, css = queue_state(q.latency) %>
+        <div class="rh-insight-row">
+          <div class="rh-insight-main"><%= link_to q.name, queues_path, class: "rh-joblink rh-mono" %><br><span class="rh-sub"><%= number_with_delimiter q.size %> waiting</span></div>
+          <span class="rh-st <%= css %>"><%= q.latency < 60 ? "#{q.latency.round(1)}s" : distance_of_time_in_words(0, q.latency) %></span>
+        </div>
+      <% end %>
+    <% end %>
+  </div>
 </div>
 
 <h2 class="rh-h2">Queues <span class="hint">live · click a queue to manage</span></h2>

data/app/views/roundhouse_ui/jobs/show.html.erb

@@ -9,7 +9,15 @@
 </div>
 
 <div class="rh-sec">Arguments<% if RoundhouseUi.redact_args.present? %> <span class="rh-sub">— sensitive keys masked</span><% end %></div>
-<pre class="rh-pre"><%= JSON.pretty_generate(RoundhouseUi::Redaction.apply(item["args"] || [])) %></pre>
+<% args_json = JSON.pretty_generate(RoundhouseUi::Redaction.apply(item["args"] || [])) %>
+<% if args_json.lines.size > 30 %>
+  <details class="rh-disclose">
+    <summary><%= pluralize(args_json.lines.size, "line") %> — expand arguments</summary>
+    <pre class="rh-pre"><%= args_json %></pre>
+  </details>
+<% else %>
+  <pre class="rh-pre"><%= args_json %></pre>
+<% end %>
 
 <% if item["error_class"].present? %>
   <div class="rh-sec">Error</div>
@@ -17,7 +25,10 @@
   <% if item["error_message"].present? %><p class="rh-sub" style="margin:6px 0 0"><%= item["error_message"] %></p><% end %>
   <% bt = item["error_backtrace"] %>
   <% if bt.is_a?(Array) && bt.any? %>
-    <pre class="rh-pre"><%= bt.first(20).join("\n") %></pre>
+    <details class="rh-disclose"<%= " open".html_safe if bt.size <= 20 %>>
+      <summary><%= pluralize(bt.size, "line") %> backtrace</summary>
+      <pre class="rh-pre"><%= bt.join("\n") %></pre>
+    </details>
   <% end %>
 <% end %>
 

data/lib/roundhouse_ui/error_groups.rb

@@ -0,0 +1,65 @@
+require "sidekiq/api"
+
+module RoundhouseUi
+  # Groups failing jobs across the retry + dead sets (and the sidekiq-failures
+  # `failed` set, when opted in) by a fingerprint of job class + error class —
+  # so one bad deploy reads as a single issue with a count, not thousands of
+  # identical rows. Used by the Errors page and the dashboard's "top failing"
+  # panel, so the aggregation lives here rather than in a controller.
+  class ErrorGroups
+    DEFAULT_SCAN_LIMIT = 1_000 # cap entries scanned per pass; surfaced honestly
+
+    Result = Struct.new(:groups, :scanned, :truncated, keyword_init: true)
+
+    def initialize(query: nil, limit: DEFAULT_SCAN_LIMIT)
+      @query = query.to_s.strip
+      @limit = limit
+    end
+
+    def call
+      groups = {}
+      scanned = 0
+      truncated = false
+
+      sources.each do |source, set|
+        set.each do |entry|
+          scanned += 1
+          if scanned > @limit
+            truncated = true
+            break
+          end
+          record(groups, source, entry)
+        end
+        break if truncated
+      end
+
+      list = groups.values.sort_by { |g| -g[:count] }
+      list = list.select { |g| "#{g[:klass]} #{g[:error]}".downcase.include?(@query.downcase) } if @query.present?
+      Result.new(groups: list, scanned: scanned, truncated: truncated)
+    end
+
+    private
+
+    # Sidekiq's native sets, plus the sidekiq-failures `failed` set when opted in
+    # and loaded. Its FailureSet is a Sidekiq::JobSet, so it iterates like the rest.
+    def sources
+      sets = { "retry" => Sidekiq::RetrySet.new, "dead" => Sidekiq::DeadSet.new }
+      if RoundhouseUi.show_sidekiq_failures && defined?(Sidekiq::Failures::FailureSet)
+        sets["failed"] = Sidekiq::Failures::FailureSet.new
+      end
+      sets
+    end
+
+    def record(groups, source, entry)
+      error = entry.item["error_class"] || "UnknownError"
+      group = (groups["#{entry.klass}|#{error}"] ||= {
+        klass: entry.klass, error: error, count: 0, last_at: nil, queues: [], sources: []
+      })
+      group[:count]  += 1
+      group[:queues] |= [ entry.queue ]
+      group[:sources] |= [ source ]
+      at = entry.at
+      group[:last_at] = at if at && (group[:last_at].nil? || at > group[:last_at])
+    end
+  end
+end

data/lib/roundhouse_ui/health.rb

@@ -0,0 +1,65 @@
+module RoundhouseUi
+  # A composite health verdict for the dashboard. Instead of a static green dot,
+  # it rolls up the signals an on-call engineer actually checks — error rate,
+  # queue latency, worker utilization — into one status + a human reason, and
+  # exposes the sub-signals so the banner can explain *why*.
+  class Health
+    Signal = Struct.new(:key, :label, :status, :detail, keyword_init: true)
+
+    RANK = { ok: 0, warn: 1, crit: 2 }.freeze
+
+    def initialize(stats:, queues:, metrics:)
+      @stats = stats
+      @queues = queues
+      @metrics = metrics
+    end
+
+    def signals
+      @signals ||= [ error_rate_signal, latency_signal, utilization_signal ].compact
+    end
+
+    # Worst sub-signal wins.
+    def status
+      signals.map(&:status).max_by { |s| RANK[s] } || :ok
+    end
+
+    def reason
+      worst = signals.max_by { |s| RANK[s.status] }
+      return "all signals nominal" if worst.nil? || worst.status == :ok
+
+      worst.detail
+    end
+
+    def healthy?
+      status == :ok
+    end
+
+    private
+
+    def error_rate_signal
+      ratio = @metrics.failure_ratio
+      status = if ratio >= 0.10 then :crit elsif ratio >= 0.02 then :warn else :ok end
+      Signal.new(key: "error_rate", label: "Error rate (lifetime)", status: status,
+                 detail: "#{(ratio * 100).round(1)}% of processed jobs have failed")
+    end
+
+    def latency_signal
+      worst = @queues.max_by(&:latency)
+      return Signal.new(key: "latency", label: "Queue latency", status: :ok, detail: "no active queues") if worst.nil?
+
+      lat = worst.latency
+      status = if lat > 600 then :crit elsif lat > 60 then :warn else :ok end
+      detail = status == :ok ? "all queues fresh (< 60s)" : "#{worst.name}: oldest job #{lat.round}s"
+      Signal.new(key: "latency", label: "Queue latency", status: status, detail: detail)
+    end
+
+    def utilization_signal
+      util = @metrics.utilization
+      return nil if util.nil? # no processes reporting in — can't judge
+
+      status = if util >= 1.0 then :crit elsif util >= 0.85 then :warn else :ok end
+      Signal.new(key: "utilization", label: "Worker utilization", status: status,
+                 detail: "#{(util * 100).round}% of worker threads busy")
+    end
+  end
+end

data/lib/roundhouse_ui/version.rb

@@ -1,3 +1,3 @@
 module RoundhouseUi
-  VERSION = "0.6.0"
+  VERSION = "0.7.0"
 end

data/lib/roundhouse_ui.rb

@@ -10,6 +10,8 @@ require "roundhouse_ui/redaction"
 require "roundhouse_ui/cancellation"
 require "roundhouse_ui/cancel_middleware"
 require "roundhouse_ui/metrics"
+require "roundhouse_ui/error_groups"
+require "roundhouse_ui/health"
 
 # Brand name is "Roundhouse"; the gem and Ruby namespace are RoundhouseUi
 # (matching the published gem name `roundhouse_ui`).

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: roundhouse_ui
 version: !ruby/object:Gem::Version
-  version: 0.6.0
+  version: 0.7.0
 platform: ruby
 authors:
 - R.J. Robinson
@@ -97,7 +97,9 @@ files:
 - lib/roundhouse_ui/cancel_middleware.rb
 - lib/roundhouse_ui/cancellation.rb
 - lib/roundhouse_ui/engine.rb
+- lib/roundhouse_ui/error_groups.rb
 - lib/roundhouse_ui/fetch.rb
+- lib/roundhouse_ui/health.rb
 - lib/roundhouse_ui/metrics.rb
 - lib/roundhouse_ui/observability.rb
 - lib/roundhouse_ui/pause.rb