From fc3ce85f8883947cf9a0c6b4700e6f5f6ffa93ef Mon Sep 17 00:00:00 2001
From: Andras Bacsai <5845193+andrasbacsai@users.noreply.github.com>
Date: Wed, 29 Apr 2026 15:40:01 +0200
Subject: [PATCH] feat(horizon): suppress failed job entries for
 deployment/timeout errors on cloud

On cloud, DeploymentException and TimeoutExceededException are expected
failure modes that pollute the Horizon failed jobs UI. Listen to JobFailed
events and scrub the entry via JobRepository::deleteFailed so operators
are not alerted for noise failures. Self-hosted instances are unaffected.
---
 app/Providers/HorizonServiceProvider.php      | 23 +++++++
 .../SuppressHorizonJobFailuresTest.php        | 65 +++++++++++++++++++
 2 files changed, 88 insertions(+)
 create mode 100644 tests/Feature/SuppressHorizonJobFailuresTest.php

diff --git a/app/Providers/HorizonServiceProvider.php b/app/Providers/HorizonServiceProvider.php
index 0caa3a3a9..c29f7fc41 100644
--- a/app/Providers/HorizonServiceProvider.php
+++ b/app/Providers/HorizonServiceProvider.php
@@ -3,9 +3,12 @@
 namespace App\Providers;
 
 use App\Contracts\CustomJobRepositoryInterface;
+use App\Exceptions\DeploymentException;
 use App\Models\ApplicationDeploymentQueue;
 use App\Models\User;
 use App\Repositories\CustomJobRepository;
+use Illuminate\Queue\Events\JobFailed;
+use Illuminate\Queue\TimeoutExceededException;
 use Illuminate\Support\Facades\Event;
 use Illuminate\Support\Facades\Gate;
 use Laravel\Horizon\Contracts\JobRepository;
@@ -48,6 +51,26 @@ public function boot(): void
                 ]);
             }
         });
+
+        Event::listen(function (JobFailed $event) {
+            if (! isCloud()) {
+                return;
+            }
+
+            $exception = $event->exception;
+            if (! ($exception instanceof DeploymentException) && ! ($exception instanceof TimeoutExceededException)) {
+                return;
+            }
+
+            try {
+                $uuid = $event->job->uuid();
+                if ($uuid) {
+                    app(JobRepository::class)->deleteFailed($uuid);
+                }
+            } catch (\Throwable $e) {
+                // Best-effort scrub; never mask the original failure.
+            }
+        });
     }
 
     protected function gate(): void
diff --git a/tests/Feature/SuppressHorizonJobFailuresTest.php b/tests/Feature/SuppressHorizonJobFailuresTest.php
new file mode 100644
index 000000000..ead342c31
--- /dev/null
+++ b/tests/Feature/SuppressHorizonJobFailuresTest.php
@@ -0,0 +1,65 @@
+<?php
+
+use App\Exceptions\DeploymentException;
+use Illuminate\Contracts\Queue\Job;
+use Illuminate\Queue\Events\JobFailed;
+use Illuminate\Queue\TimeoutExceededException;
+use Laravel\Horizon\Contracts\JobRepository;
+use Mockery\MockInterface;
+
+function fakeJob(string $uuid): Job
+{
+    $job = Mockery::mock(Job::class)->shouldIgnoreMissing();
+    $job->shouldReceive('uuid')->andReturn($uuid);
+    $job->shouldReceive('getJobId')->andReturn($uuid);
+
+    return $job;
+}
+
+function fireJobFailed(Job $job, Throwable $exception): void
+{
+    event(new JobFailed('redis', $job, $exception));
+}
+
+beforeEach(function () {
+    config(['constants.coolify.self_hosted' => false]);
+});
+
+test('scrubs Horizon failed entry for DeploymentException on cloud', function () {
+    $uuid = 'uuid-deployment-1';
+
+    $this->mock(JobRepository::class, function (MockInterface $mock) use ($uuid) {
+        $mock->shouldReceive('deleteFailed')->once()->with($uuid);
+    });
+
+    fireJobFailed(fakeJob($uuid), new DeploymentException('build failed'));
+});
+
+test('scrubs Horizon failed entry for TimeoutExceededException on cloud', function () {
+    $uuid = 'uuid-timeout-1';
+
+    $this->mock(JobRepository::class, function (MockInterface $mock) use ($uuid) {
+        $mock->shouldReceive('deleteFailed')->once()->with($uuid);
+    });
+
+    fireJobFailed(fakeJob($uuid), new TimeoutExceededException('worker timeout'));
+});
+
+test('does not scrub generic exceptions on cloud', function () {
+    $this->mock(JobRepository::class, function (MockInterface $mock) {
+        $mock->shouldNotReceive('deleteFailed');
+    });
+
+    fireJobFailed(fakeJob('uuid-generic-1'), new RuntimeException('boom'));
+});
+
+test('does not scrub when self-hosted even for filtered exceptions', function () {
+    config(['constants.coolify.self_hosted' => true]);
+
+    $this->mock(JobRepository::class, function (MockInterface $mock) {
+        $mock->shouldNotReceive('deleteFailed');
+    });
+
+    fireJobFailed(fakeJob('uuid-deployment-2'), new DeploymentException('build failed'));
+    fireJobFailed(fakeJob('uuid-timeout-2'), new TimeoutExceededException('worker timeout'));
+});