mirror of
https://github.com/tiennm99/coolify.git
synced 2026-10-15 00:17:16 +00:00
feat(jobs): add queue delay resilience to scheduled job execution
Implement dedup key-based cron tracking to make scheduled jobs resilient to queue delays. Even if a job is delayed by minutes, it will catch the missed cron window by tracking previousRunDate in cache instead of relying on isDue() alone. - Add dedupKey parameter to shouldRunNow() in ScheduledJobManager - When provided, uses getPreviousRunDate() + cache tracking for resilience - Falls back to isDue() for docker cleanups without dedup key - Prevents double-dispatch within same cron window - Optimize ServerConnectionCheckJob dispatch - Skip SSH checks if Sentinel is healthy (enabled and live) - Reduces redundant checks when Sentinel heartbeat proves connectivity - Remove hourly Sentinel update checks - Consolidate to daily CheckAndStartSentinelJob dispatch - Crash recovery handled by sentinelOutOfSync → ServerCheckJob flow - Add logging for skipped database backups with context (backup_id, database_id, status) - Refactor skip reason methods to accept server parameter, avoiding redundant queries - Add comprehensive test suite for scheduling with various delay scenarios and timezones
This commit is contained in:
1 parent
f68793ed69
commit
a0c177f6f2
5 files changed
+322
-106
No files matched your search
@@ -0,0 +1,208 @@
|
||||
<?php
|
||||
|
||||
use App\Jobs\ScheduledJobManager;
|
||||
use Illuminate\Support\Carbon;
|
||||
use Illuminate\Support\Facades\Cache;
|
||||
|
||||
beforeEach(function () {
|
||||
// Clear any dedup keys
|
||||
Cache::flush();
|
||||
});
|
||||
|
||||
it('dispatches backup when job runs on time at the cron minute', function () {
|
||||
// Freeze time at exactly 02:00 — daily cron "0 2 * * *" is due
|
||||
Carbon::setTestNow(Carbon::create(2026, 2, 28, 2, 0, 0, 'UTC'));
|
||||
|
||||
$job = new ScheduledJobManager;
|
||||
|
||||
// Use reflection to test shouldRunNow
|
||||
$reflection = new ReflectionClass($job);
|
||||
|
||||
$executionTimeProp = $reflection->getProperty('executionTime');
|
||||
$executionTimeProp->setAccessible(true);
|
||||
$executionTimeProp->setValue($job, Carbon::now());
|
||||
|
||||
$method = $reflection->getMethod('shouldRunNow');
|
||||
$method->setAccessible(true);
|
||||
|
||||
$result = $method->invoke($job, '0 2 * * *', 'UTC', 'test-backup:1');
|
||||
|
||||
expect($result)->toBeTrue();
|
||||
});
|
||||
|
||||
it('dispatches backup when job is delayed past the cron minute', function () {
|
||||
// Freeze time at 02:07 — job was delayed 7 minutes past the 02:00 cron
|
||||
Carbon::setTestNow(Carbon::create(2026, 2, 28, 2, 7, 0, 'UTC'));
|
||||
|
||||
$job = new ScheduledJobManager;
|
||||
|
||||
$reflection = new ReflectionClass($job);
|
||||
|
||||
$executionTimeProp = $reflection->getProperty('executionTime');
|
||||
$executionTimeProp->setAccessible(true);
|
||||
$executionTimeProp->setValue($job, Carbon::now());
|
||||
|
||||
$method = $reflection->getMethod('shouldRunNow');
|
||||
$method->setAccessible(true);
|
||||
|
||||
// isDue() would return false at 02:07, but getPreviousRunDate() = 02:00
|
||||
// No lastDispatched in cache → should run
|
||||
$result = $method->invoke($job, '0 2 * * *', 'UTC', 'test-backup:1');
|
||||
|
||||
expect($result)->toBeTrue();
|
||||
});
|
||||
|
||||
it('does not double-dispatch on subsequent runs within same cron window', function () {
|
||||
// First run at 02:00 — dispatches and sets cache
|
||||
Carbon::setTestNow(Carbon::create(2026, 2, 28, 2, 0, 0, 'UTC'));
|
||||
|
||||
$job = new ScheduledJobManager;
|
||||
$reflection = new ReflectionClass($job);
|
||||
|
||||
$executionTimeProp = $reflection->getProperty('executionTime');
|
||||
$executionTimeProp->setAccessible(true);
|
||||
$executionTimeProp->setValue($job, Carbon::now());
|
||||
|
||||
$method = $reflection->getMethod('shouldRunNow');
|
||||
$method->setAccessible(true);
|
||||
|
||||
$first = $method->invoke($job, '0 2 * * *', 'UTC', 'test-backup:2');
|
||||
expect($first)->toBeTrue();
|
||||
|
||||
// Second run at 02:01 — should NOT dispatch (previousDue=02:00, lastDispatched=02:00)
|
||||
Carbon::setTestNow(Carbon::create(2026, 2, 28, 2, 1, 0, 'UTC'));
|
||||
$executionTimeProp->setValue($job, Carbon::now());
|
||||
|
||||
$second = $method->invoke($job, '0 2 * * *', 'UTC', 'test-backup:2');
|
||||
expect($second)->toBeFalse();
|
||||
});
|
||||
|
||||
it('fires every_minute cron correctly on consecutive minutes', function () {
|
||||
$job = new ScheduledJobManager;
|
||||
$reflection = new ReflectionClass($job);
|
||||
|
||||
$executionTimeProp = $reflection->getProperty('executionTime');
|
||||
$executionTimeProp->setAccessible(true);
|
||||
|
||||
$method = $reflection->getMethod('shouldRunNow');
|
||||
$method->setAccessible(true);
|
||||
|
||||
// Minute 1
|
||||
Carbon::setTestNow(Carbon::create(2026, 2, 28, 10, 0, 0, 'UTC'));
|
||||
$executionTimeProp->setValue($job, Carbon::now());
|
||||
$result1 = $method->invoke($job, '* * * * *', 'UTC', 'test-backup:3');
|
||||
expect($result1)->toBeTrue();
|
||||
|
||||
// Minute 2
|
||||
Carbon::setTestNow(Carbon::create(2026, 2, 28, 10, 1, 0, 'UTC'));
|
||||
$executionTimeProp->setValue($job, Carbon::now());
|
||||
$result2 = $method->invoke($job, '* * * * *', 'UTC', 'test-backup:3');
|
||||
expect($result2)->toBeTrue();
|
||||
|
||||
// Minute 3
|
||||
Carbon::setTestNow(Carbon::create(2026, 2, 28, 10, 2, 0, 'UTC'));
|
||||
$executionTimeProp->setValue($job, Carbon::now());
|
||||
$result3 = $method->invoke($job, '* * * * *', 'UTC', 'test-backup:3');
|
||||
expect($result3)->toBeTrue();
|
||||
});
|
||||
|
||||
it('re-dispatches after cache loss instead of missing', function () {
|
||||
// First run at 02:00 — dispatches
|
||||
Carbon::setTestNow(Carbon::create(2026, 2, 28, 2, 0, 0, 'UTC'));
|
||||
|
||||
$job = new ScheduledJobManager;
|
||||
$reflection = new ReflectionClass($job);
|
||||
|
||||
$executionTimeProp = $reflection->getProperty('executionTime');
|
||||
$executionTimeProp->setAccessible(true);
|
||||
$executionTimeProp->setValue($job, Carbon::now());
|
||||
|
||||
$method = $reflection->getMethod('shouldRunNow');
|
||||
$method->setAccessible(true);
|
||||
|
||||
$method->invoke($job, '0 2 * * *', 'UTC', 'test-backup:4');
|
||||
|
||||
// Simulate Redis restart — cache lost
|
||||
Cache::forget('test-backup:4');
|
||||
|
||||
// Run again at 02:01 — should re-dispatch (lastDispatched = null after cache loss)
|
||||
Carbon::setTestNow(Carbon::create(2026, 2, 28, 2, 1, 0, 'UTC'));
|
||||
$executionTimeProp->setValue($job, Carbon::now());
|
||||
|
||||
$result = $method->invoke($job, '0 2 * * *', 'UTC', 'test-backup:4');
|
||||
expect($result)->toBeTrue();
|
||||
});
|
||||
|
||||
it('does not dispatch when cron is not due and was not recently due', function () {
|
||||
// Time is 10:00, cron is daily at 02:00 — last due was 8 hours ago
|
||||
Carbon::setTestNow(Carbon::create(2026, 2, 28, 10, 0, 0, 'UTC'));
|
||||
|
||||
$job = new ScheduledJobManager;
|
||||
$reflection = new ReflectionClass($job);
|
||||
|
||||
$executionTimeProp = $reflection->getProperty('executionTime');
|
||||
$executionTimeProp->setAccessible(true);
|
||||
$executionTimeProp->setValue($job, Carbon::now());
|
||||
|
||||
$method = $reflection->getMethod('shouldRunNow');
|
||||
$method->setAccessible(true);
|
||||
|
||||
// previousDue = 02:00, but lastDispatched was set at 02:00 (simulate)
|
||||
Cache::put('test-backup:5', Carbon::create(2026, 2, 28, 2, 0, 0, 'UTC')->toIso8601String(), 86400);
|
||||
|
||||
$result = $method->invoke($job, '0 2 * * *', 'UTC', 'test-backup:5');
|
||||
expect($result)->toBeFalse();
|
||||
});
|
||||
|
||||
it('falls back to isDue when no dedup key is provided', function () {
|
||||
// Time is exactly 02:00, cron is "0 2 * * *" — should be due
|
||||
Carbon::setTestNow(Carbon::create(2026, 2, 28, 2, 0, 0, 'UTC'));
|
||||
|
||||
$job = new ScheduledJobManager;
|
||||
$reflection = new ReflectionClass($job);
|
||||
|
||||
$executionTimeProp = $reflection->getProperty('executionTime');
|
||||
$executionTimeProp->setAccessible(true);
|
||||
$executionTimeProp->setValue($job, Carbon::now());
|
||||
|
||||
$method = $reflection->getMethod('shouldRunNow');
|
||||
$method->setAccessible(true);
|
||||
|
||||
// No dedup key → simple isDue check
|
||||
$result = $method->invoke($job, '0 2 * * *', 'UTC');
|
||||
expect($result)->toBeTrue();
|
||||
|
||||
// At 02:01 without dedup key → isDue returns false
|
||||
Carbon::setTestNow(Carbon::create(2026, 2, 28, 2, 1, 0, 'UTC'));
|
||||
$executionTimeProp->setValue($job, Carbon::now());
|
||||
|
||||
$result2 = $method->invoke($job, '0 2 * * *', 'UTC');
|
||||
expect($result2)->toBeFalse();
|
||||
});
|
||||
|
||||
it('respects server timezone for cron evaluation', function () {
|
||||
// UTC time is 22:00 Feb 28, which is 06:00 Mar 1 in Asia/Singapore (+8)
|
||||
Carbon::setTestNow(Carbon::create(2026, 2, 28, 22, 0, 0, 'UTC'));
|
||||
|
||||
$job = new ScheduledJobManager;
|
||||
$reflection = new ReflectionClass($job);
|
||||
|
||||
$executionTimeProp = $reflection->getProperty('executionTime');
|
||||
$executionTimeProp->setAccessible(true);
|
||||
$executionTimeProp->setValue($job, Carbon::now());
|
||||
|
||||
$method = $reflection->getMethod('shouldRunNow');
|
||||
$method->setAccessible(true);
|
||||
|
||||
// Simulate that today's 06:00 UTC run was already dispatched (at 06:00 UTC)
|
||||
Cache::put('test-backup:7', Carbon::create(2026, 2, 28, 6, 0, 0, 'UTC')->toIso8601String(), 86400);
|
||||
|
||||
// Cron "0 6 * * *" in Asia/Singapore: local time is 06:00 Mar 1 → previousDue = 06:00 Mar 1 SGT
|
||||
// That's a NEW cron window (Mar 1) that hasn't been dispatched → should fire
|
||||
$resultSingapore = $method->invoke($job, '0 6 * * *', 'Asia/Singapore', 'test-backup:6');
|
||||
expect($resultSingapore)->toBeTrue();
|
||||
|
||||
// Cron "0 6 * * *" in UTC: previousDue = 06:00 Feb 28 UTC, already dispatched at 06:00 → should NOT fire
|
||||
$resultUtc = $method->invoke($job, '0 6 * * *', 'UTC', 'test-backup:7');
|
||||
expect($resultUtc)->toBeFalse();
|
||||
});
|
||||
@@ -1,6 +1,7 @@
|
||||
<?php
|
||||
|
||||
use App\Jobs\CheckAndStartSentinelJob;
|
||||
use App\Jobs\ServerConnectionCheckJob;
|
||||
use App\Jobs\ServerManagerJob;
|
||||
use App\Models\InstanceSettings;
|
||||
use App\Models\Server;
|
||||
@@ -10,23 +11,22 @@ use Mockery;
|
||||
|
||||
beforeEach(function () {
|
||||
Queue::fake();
|
||||
Carbon::setTestNow('2025-01-15 12:00:00'); // Set to top of the hour for cron matching
|
||||
Carbon::setTestNow('2025-01-15 12:00:00');
|
||||
});
|
||||
|
||||
afterEach(function () {
|
||||
Mockery::close();
|
||||
Carbon::setTestNow(); // Reset frozen time
|
||||
Carbon::setTestNow();
|
||||
});
|
||||
|
||||
it('dispatches CheckAndStartSentinelJob hourly for sentinel-enabled servers', function () {
|
||||
// Mock InstanceSettings
|
||||
it('does not dispatch CheckAndStartSentinelJob hourly anymore', function () {
|
||||
$settings = Mockery::mock(InstanceSettings::class);
|
||||
$settings->instance_timezone = 'UTC';
|
||||
$this->app->instance(InstanceSettings::class, $settings);
|
||||
|
||||
// Create a mock server with sentinel enabled
|
||||
$server = Mockery::mock(Server::class)->makePartial();
|
||||
$server->shouldReceive('isSentinelEnabled')->andReturn(true);
|
||||
$server->shouldReceive('isSentinelLive')->andReturn(true);
|
||||
$server->id = 1;
|
||||
$server->name = 'test-server';
|
||||
$server->ip = '192.168.1.100';
|
||||
@@ -34,29 +34,76 @@ it('dispatches CheckAndStartSentinelJob hourly for sentinel-enabled servers', fu
|
||||
$server->shouldReceive('getAttribute')->with('settings')->andReturn((object) ['server_timezone' => 'UTC']);
|
||||
$server->shouldReceive('waitBeforeDoingSshCheck')->andReturn(120);
|
||||
|
||||
// Mock the Server query
|
||||
Server::shouldReceive('where')->with('ip', '!=', '1.2.3.4')->andReturnSelf();
|
||||
Server::shouldReceive('get')->andReturn(collect([$server]));
|
||||
|
||||
// Execute the job
|
||||
$job = new ServerManagerJob;
|
||||
$job->handle();
|
||||
|
||||
// Assert CheckAndStartSentinelJob was dispatched for the sentinel-enabled server
|
||||
Queue::assertPushed(CheckAndStartSentinelJob::class, function ($job) use ($server) {
|
||||
return $job->server->id === $server->id;
|
||||
});
|
||||
// Hourly CheckAndStartSentinelJob dispatch was removed — ServerCheckJob handles it when Sentinel is out of sync
|
||||
Queue::assertNotPushed(CheckAndStartSentinelJob::class);
|
||||
});
|
||||
|
||||
it('does not dispatch CheckAndStartSentinelJob for servers without sentinel enabled', function () {
|
||||
// Mock InstanceSettings
|
||||
it('skips ServerConnectionCheckJob when sentinel is live', function () {
|
||||
$settings = Mockery::mock(InstanceSettings::class);
|
||||
$settings->instance_timezone = 'UTC';
|
||||
$this->app->instance(InstanceSettings::class, $settings);
|
||||
|
||||
$server = Mockery::mock(Server::class)->makePartial();
|
||||
$server->shouldReceive('isSentinelEnabled')->andReturn(true);
|
||||
$server->shouldReceive('isSentinelLive')->andReturn(true);
|
||||
$server->id = 1;
|
||||
$server->name = 'test-server';
|
||||
$server->ip = '192.168.1.100';
|
||||
$server->sentinel_updated_at = Carbon::now();
|
||||
$server->shouldReceive('getAttribute')->with('settings')->andReturn((object) ['server_timezone' => 'UTC']);
|
||||
$server->shouldReceive('waitBeforeDoingSshCheck')->andReturn(120);
|
||||
|
||||
Server::shouldReceive('where')->with('ip', '!=', '1.2.3.4')->andReturnSelf();
|
||||
Server::shouldReceive('get')->andReturn(collect([$server]));
|
||||
|
||||
$job = new ServerManagerJob;
|
||||
$job->handle();
|
||||
|
||||
// Sentinel is healthy so SSH connection check is skipped
|
||||
Queue::assertNotPushed(ServerConnectionCheckJob::class);
|
||||
});
|
||||
|
||||
it('dispatches ServerConnectionCheckJob when sentinel is not live', function () {
|
||||
$settings = Mockery::mock(InstanceSettings::class);
|
||||
$settings->instance_timezone = 'UTC';
|
||||
$this->app->instance(InstanceSettings::class, $settings);
|
||||
|
||||
$server = Mockery::mock(Server::class)->makePartial();
|
||||
$server->shouldReceive('isSentinelEnabled')->andReturn(true);
|
||||
$server->shouldReceive('isSentinelLive')->andReturn(false);
|
||||
$server->id = 1;
|
||||
$server->name = 'test-server';
|
||||
$server->ip = '192.168.1.100';
|
||||
$server->sentinel_updated_at = Carbon::now()->subMinutes(10);
|
||||
$server->shouldReceive('getAttribute')->with('settings')->andReturn((object) ['server_timezone' => 'UTC']);
|
||||
$server->shouldReceive('waitBeforeDoingSshCheck')->andReturn(120);
|
||||
|
||||
Server::shouldReceive('where')->with('ip', '!=', '1.2.3.4')->andReturnSelf();
|
||||
Server::shouldReceive('get')->andReturn(collect([$server]));
|
||||
|
||||
$job = new ServerManagerJob;
|
||||
$job->handle();
|
||||
|
||||
// Sentinel is out of sync so SSH connection check is needed
|
||||
Queue::assertPushed(ServerConnectionCheckJob::class, function ($job) use ($server) {
|
||||
return $job->server->id === $server->id;
|
||||
});
|
||||
});
|
||||
|
||||
it('dispatches ServerConnectionCheckJob when sentinel is not enabled', function () {
|
||||
$settings = Mockery::mock(InstanceSettings::class);
|
||||
$settings->instance_timezone = 'UTC';
|
||||
$this->app->instance(InstanceSettings::class, $settings);
|
||||
|
||||
// Create a mock server with sentinel disabled
|
||||
$server = Mockery::mock(Server::class)->makePartial();
|
||||
$server->shouldReceive('isSentinelEnabled')->andReturn(false);
|
||||
$server->shouldReceive('isSentinelLive')->never();
|
||||
$server->id = 2;
|
||||
$server->name = 'test-server-no-sentinel';
|
||||
$server->ip = '192.168.1.101';
|
||||
@@ -64,78 +111,14 @@ it('does not dispatch CheckAndStartSentinelJob for servers without sentinel enab
|
||||
$server->shouldReceive('getAttribute')->with('settings')->andReturn((object) ['server_timezone' => 'UTC']);
|
||||
$server->shouldReceive('waitBeforeDoingSshCheck')->andReturn(120);
|
||||
|
||||
// Mock the Server query
|
||||
Server::shouldReceive('where')->with('ip', '!=', '1.2.3.4')->andReturnSelf();
|
||||
Server::shouldReceive('get')->andReturn(collect([$server]));
|
||||
|
||||
// Execute the job
|
||||
$job = new ServerManagerJob;
|
||||
$job->handle();
|
||||
|
||||
// Assert CheckAndStartSentinelJob was NOT dispatched
|
||||
Queue::assertNotPushed(CheckAndStartSentinelJob::class);
|
||||
});
|
||||
|
||||
it('respects server timezone when scheduling sentinel checks', function () {
|
||||
// Mock InstanceSettings
|
||||
$settings = Mockery::mock(InstanceSettings::class);
|
||||
$settings->instance_timezone = 'UTC';
|
||||
$this->app->instance(InstanceSettings::class, $settings);
|
||||
|
||||
// Set test time to top of hour in America/New_York (which is 17:00 UTC)
|
||||
Carbon::setTestNow('2025-01-15 17:00:00'); // 12:00 PM EST (top of hour in EST)
|
||||
|
||||
// Create a mock server with sentinel enabled and America/New_York timezone
|
||||
$server = Mockery::mock(Server::class)->makePartial();
|
||||
$server->shouldReceive('isSentinelEnabled')->andReturn(true);
|
||||
$server->id = 3;
|
||||
$server->name = 'test-server-est';
|
||||
$server->ip = '192.168.1.102';
|
||||
$server->sentinel_updated_at = Carbon::now();
|
||||
$server->shouldReceive('getAttribute')->with('settings')->andReturn((object) ['server_timezone' => 'America/New_York']);
|
||||
$server->shouldReceive('waitBeforeDoingSshCheck')->andReturn(120);
|
||||
|
||||
// Mock the Server query
|
||||
Server::shouldReceive('where')->with('ip', '!=', '1.2.3.4')->andReturnSelf();
|
||||
Server::shouldReceive('get')->andReturn(collect([$server]));
|
||||
|
||||
// Execute the job
|
||||
$job = new ServerManagerJob;
|
||||
$job->handle();
|
||||
|
||||
// Assert CheckAndStartSentinelJob was dispatched (should run at top of hour in server's timezone)
|
||||
Queue::assertPushed(CheckAndStartSentinelJob::class, function ($job) use ($server) {
|
||||
// Sentinel is not enabled so SSH connection check must run
|
||||
Queue::assertPushed(ServerConnectionCheckJob::class, function ($job) use ($server) {
|
||||
return $job->server->id === $server->id;
|
||||
});
|
||||
});
|
||||
|
||||
it('does not dispatch sentinel check when not at top of hour', function () {
|
||||
// Mock InstanceSettings
|
||||
$settings = Mockery::mock(InstanceSettings::class);
|
||||
$settings->instance_timezone = 'UTC';
|
||||
$this->app->instance(InstanceSettings::class, $settings);
|
||||
|
||||
// Set test time to middle of the hour (not top of hour)
|
||||
Carbon::setTestNow('2025-01-15 12:30:00');
|
||||
|
||||
// Create a mock server with sentinel enabled
|
||||
$server = Mockery::mock(Server::class)->makePartial();
|
||||
$server->shouldReceive('isSentinelEnabled')->andReturn(true);
|
||||
$server->id = 4;
|
||||
$server->name = 'test-server-mid-hour';
|
||||
$server->ip = '192.168.1.103';
|
||||
$server->sentinel_updated_at = Carbon::now();
|
||||
$server->shouldReceive('getAttribute')->with('settings')->andReturn((object) ['server_timezone' => 'UTC']);
|
||||
$server->shouldReceive('waitBeforeDoingSshCheck')->andReturn(120);
|
||||
|
||||
// Mock the Server query
|
||||
Server::shouldReceive('where')->with('ip', '!=', '1.2.3.4')->andReturnSelf();
|
||||
Server::shouldReceive('get')->andReturn(collect([$server]));
|
||||
|
||||
// Execute the job
|
||||
$job = new ServerManagerJob;
|
||||
$job->handle();
|
||||
|
||||
// Assert CheckAndStartSentinelJob was NOT dispatched (not top of hour)
|
||||
Queue::assertNotPushed(CheckAndStartSentinelJob::class);
|
||||
});
|
||||
Reference in new issue
Block a user