feat: failure rate sparkline per queue + QueueStats service object

CHANGELOG.md

@@ -7,6 +7,10 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+### Added
+
+- Failure rate sparkline per queue — a mini 12-bar chart in the Queues table shows the percentage of jobs that failed (vs. completed) in each of the last 12 hours; bars are red, sized proportionally to the failure rate (0–100 %), and include a tooltip with the hour label and exact percentage; empty hours render as a faint border-colored bar; queues with no activity in the last 12 hours show "—"
+
 ## [0.8.0] - 2026-05-20
 
 ### Added

README.md

@@ -34,7 +34,7 @@ SolidQueueWeb surfaces all of this in a browser UI available at any route you ch
 ## Features
 
 - **Dashboard** — stat cards showing counts for ready, scheduled, running, blocked, and failed jobs, plus queues, recurring tasks, and processes; "Done (1h)" and "Done (24h)" throughput cards; a "Throughput — Last 12 Hours" bar chart showing completed-job counts per hour (pure CSS, no charting library); auto-refreshes every 5 seconds
-- **Queues** — all queues sorted by name with size; oldest ready job latency (color-coded, with UTC timestamp tooltip); Done (24h) and Failed (24h) throughput counts; pause/resume controls
+- **Queues** — all queues sorted by name with size; oldest ready job latency (color-coded, with UTC timestamp tooltip); Done (24h) and Failed (24h) throughput counts; a mini 12-bar failure rate sparkline per queue showing failure % per hour over the last 12 hours; pause/resume controls
 - **Jobs** — filterable by status (ready, scheduled, claimed, blocked, failed) and by queue; search by job class name with dynamic auto-submit; time-based period filter (1 h / 24 h / 7 d); discard individual or all jobs; Turbo Frame navigation so only the table updates on filter or search; auto-refreshes every 10 seconds
 - **Failed jobs** — list of failed executions with error details; search by class name; filter by queue; time-based period filter; retry or discard individually or in bulk
 - **Job detail** — full arguments, timestamps, blocked-until date, and error backtrace; action buttons based on job status
@@ -112,7 +112,6 @@ No authentication is enforced by default. When the `authenticate` block returns
 Planned features, roughly ordered by priority:  
 
 **Observability**
-- Job failure rate chart — sparkline per queue showing failure percentage over time, mirroring the throughput chart
 - Queue depth trend — historical queue size over time, not just the current snapshot
 - Slow job detection — flag jobs exceeding a configurable duration threshold
 

app/assets/stylesheets/solid_queue_web/_11_throughput.css

@@ -65,4 +65,29 @@
   padding: 1rem 1.25rem;
 }
 
-.sqd-stat--done .sqd-stat__value { color: var(--success); }
+.sqd-stat--done .sqd-stat__value { color: var(--success); }
+
+.sqd-mini-sparkline {
+  display: flex;
+  align-items: flex-end;
+  gap: 2px;
+  height: 28px;
+  width: 88px;
+}
+
+.sqd-mini-sparkline__bar {
+  flex: 1;
+  background: var(--danger);
+  border-radius: 1px 1px 0 0;
+  opacity: 0.7;
+  transition: opacity 0.15s;
+}
+
+.sqd-mini-sparkline__bar:hover {
+  opacity: 1;
+}
+
+.sqd-mini-sparkline__bar--empty {
+  background: var(--border);
+  opacity: 0.5;
+}

app/controllers/solid_queue_web/queues_controller.rb

@@ -2,21 +2,11 @@ module SolidQueueWeb
   class QueuesController < ApplicationController
     def index
       @queues = SolidQueue::Queue.all.sort_by(&:name)
-
-      now = Time.current
-      @completed_24h = SolidQueue::Job
-        .where(finished_at: 24.hours.ago..now)
-        .group(:queue_name)
-        .count
-      @failed_24h = SolidQueue::FailedExecution
-        .joins(:job)
-        .where(created_at: 24.hours.ago..now)
-        .group("solid_queue_jobs.queue_name")
-        .count
-      @oldest_ready = SolidQueue::ReadyExecution
-        .joins(:job)
-        .group("solid_queue_jobs.queue_name")
-        .minimum("solid_queue_jobs.created_at")
+      stats = QueueStats.new(@queues)
+      @completed_24h      = stats.completed_24h
+      @failed_24h         = stats.failed_24h
+      @oldest_ready       = stats.oldest_ready
+      @failure_sparklines = stats.failure_sparklines
     end
 
     def pause

app/services/solid_queue_web/queue_stats.rb

@@ -0,0 +1,52 @@
+module SolidQueueWeb
+  class QueueStats
+    attr_reader :completed_24h, :failed_24h, :oldest_ready, :failure_sparklines
+
+    def initialize(queues)
+      @queues = queues
+      @now    = Time.current
+      compute
+    end
+
+    private
+
+    def compute
+      @completed_24h = SolidQueue::Job
+        .where(finished_at: 24.hours.ago..@now)
+        .group(:queue_name)
+        .count
+
+      @failed_24h = SolidQueue::FailedExecution
+        .joins(:job)
+        .where(created_at: 24.hours.ago..@now)
+        .group("solid_queue_jobs.queue_name")
+        .count
+
+      @oldest_ready = SolidQueue::ReadyExecution
+        .joins(:job)
+        .group("solid_queue_jobs.queue_name")
+        .minimum("solid_queue_jobs.created_at")
+
+      failed_raw = SolidQueue::FailedExecution
+        .joins(:job)
+        .where(created_at: 12.hours.ago..@now)
+        .pluck("solid_queue_jobs.queue_name", "solid_queue_failed_executions.created_at")
+      done_raw = SolidQueue::Job
+        .where(finished_at: 12.hours.ago..@now)
+        .pluck(:queue_name, :finished_at)
+
+      @failure_sparklines = @queues.each_with_object({}) do |queue, h|
+        failed_times = failed_raw.filter_map { |q, t| t if q == queue.name }
+        done_times   = done_raw.filter_map   { |q, t| t if q == queue.name }
+        h[queue.name] = 12.times.map do |i|
+          from = (12 - i).hours.ago
+          to   = i == 11 ? @now : (11 - i).hours.ago
+          f = failed_times.count { |t| t >= from && t < to }
+          d = done_times.count   { |t| t >= from && t < to }
+          total = f + d
+          total > 0 ? (f.to_f / total * 100).round : nil
+        end
+      end
+    end
+  end
+end

app/views/solid_queue_web/queues/index.html.erb

@@ -12,6 +12,7 @@
           <th scope="col">Latency</th>
           <th scope="col">Done (24h)</th>
           <th scope="col">Failed (24h)</th>
+          <th scope="col">Failure Rate (12h)</th>
           <th scope="col">Status</th>
           <th scope="col"><span class="sqd-sr-only">Actions</span></th>
         </tr>
@@ -34,6 +35,22 @@
             </td>
             <td style="color: var(--success);"><%= @completed_24h[queue.name] || 0 %></td>
             <td style="color: <%= (@failed_24h[queue.name] || 0) > 0 ? "var(--danger)" : "inherit" %>;"><%= @failed_24h[queue.name] || 0 %></td>
+            <td>
+              <% sparkline = @failure_sparklines[queue.name] %>
+              <% if sparkline.any? %>
+                <div class="sqd-mini-sparkline" aria-label="Failure rate last 12 hours for <%= queue.name %>">
+                  <% sparkline.each_with_index do |rate, i| %>
+                    <% pct = rate || 0 %>
+                    <% hour_label = (12 - i).hours.ago.strftime("%-I%p").downcase %>
+                    <div class="sqd-mini-sparkline__bar sqd-mini-sparkline__bar--<%= rate ? "data" : "empty" %>"
+                         style="height: <%= [pct, 2].max %>%"
+                         title="<%= hour_label %>: <%= rate ? "#{rate}% failure rate" : "no data" %>"></div>
+                  <% end %>
+                </div>
+              <% else %>
+                <span style="color: var(--muted)">—</span>
+              <% end %>
+            </td>
             <td>
               <% if queue.paused? %>
                 <span class="sqd-badge sqd-badge--paused">Paused</span>

spec/dummy/db/seeds.rb

@@ -106,6 +106,26 @@
   )
 end
 
+puts "Seeding recent failed jobs (for failure rate sparkline)..."
+3.times do |i|
+  err = errors[i % errors.size]
+  job = SolidQueue::Job.create!(
+    queue_name: queues.sample,
+    class_name: job_classes.sample,
+    arguments: { recent_fail: true, idx: i }.to_json,
+    priority: 0,
+    active_job_id: SecureRandom.uuid,
+    created_at: rand(1..10).hours.ago,
+    updated_at: rand(1..10).hours.ago
+  )
+  job.ready_execution&.destroy
+  SolidQueue::FailedExecution.create!(
+    job: job,
+    error: { exception_class: err[:class], message: err[:message], backtrace: ["app/jobs/#{job.class_name.underscore}.rb:42"] },
+    created_at: job.created_at
+  )
+end
+
 puts "Seeding blocked jobs..."
 # Skip set_expires_at callback which requires a real ActiveJob class to exist
 SolidQueue::BlockedExecution.skip_callback(:create, :before, :set_expires_at)