Laravel批量任务失败后添加新任务无法执行求助

问题背景 #

我尝试了多种方法仍无法解决该问题，且确认最近一次框架更新前功能完全正常（对比源码未找到相关线索）。

需求 #

当批量中的StartTestJob耗尽重试次数失败后，向该批量中添加并执行FindNextSiteTestJob，此任务负责查找未处理的站点，并向批量中添加新的StartTestJob。

环境说明 #

• 后台Worker持续运行（通过pgrep监控）
• 使用mateusjunges/trackable-jobs-for-laravel包跟踪批量任务
• 采用数据库驱动队列
• 代码修改后会执行php artisan queue:restart

当前问题 #

FindNextSiteTestJob已被正确加入批量，但整个批量被标记为失败，该任务一直处于排队状态，批量执行完全停滞。

启动批量任务代码 #

$server = Server::find($server_id);
$sites = $server->sites->all();
$max_simultaneous_procs = $server->simultaneous_procs;

$batchJobs = [];

$batch_reference_uuid = (string) Str::orderedUuid();
$sites = array_slice($sites, 0, $max_simultaneous_procs);

foreach ($sites as $site) {
  $chain_uuid = (string) Str::orderedUuid();
  $batchJobs [] = [new StartTestJob($site, $chain_uuid, $batch_reference_uuid)];
}

/**********************
// 分发批量任务
**********************/
$queue_name = 'procs_test_server_' . $server_id;
$batch_name = 'server-'. $server_id . ' : TestBatchSimultaneous' . ' ' . Carbon::now()->format('Y-m-d H:i');
$log_path = 'uuuuuu/storage/logs/jobs.log';

try {
    $batch = Bus::batch(
        $batchJobs
    )->then(...)->catch(...)->finally(...)
    ->onQueue($queue_name)->name($batch_name)->withOption("reference_uuid", $batch_reference_uuid)->dispatch();
    // ->allowFailures() 没有任何效果
}

队列停滞状态 #

StartTestJob执行失败后，FindNextSiteTestJob进入队列但永远不会被执行。

相关任务代码 #

StartTestJob #

class StartTestJob implements ShouldQueue
{
    use Dispatchable, InteractsWithQueue, Queueable, SerializesModels, Batchable, ChainedTrackable {
        __construct as __baseConstruct;
    }
    public $site; public $chain_uuid; public $batch_reference_uuid;

    public function __construct(Site $site, string $chain_uuid, string $batch_reference_uuid)
    {
        $this->__baseConstruct($site, $chain_uuid, $batch_reference_uuid);
        $this->site = $site; $this->chain_uuid = $chain_uuid; $this->batch_reference_uuid = $batch_reference_uuid;
    }

    public function handle()
    {       
        // 测试环境中此处查询结果永远为空
        $pingback_happened = Pingback::where(['chain_uuid' => $this->chain_uuid])->first();
        if(!$pingback_happened ){ $this->release(5); return; }
    }

    public function retryUntil(){ return now()->addSeconds(15);}

    public function failed($exception)
    {
        new FindNextSiteTestJob($this->site, $this->chain_uuid, $this->batch_reference_uuid);
    }
}

FindNextSiteTestJob #

class FindNextSiteTestJob implements ShouldQueue, ShouldBeUnique
{
   use Dispatchable, InteractsWithQueue, Queueable, SerializesModels, Batchable, ChainedTrackable {
        __construct as __baseConstruct;
    }

    public $site; public $chain_uuid; public $batch_reference_uuid;

    public function __construct(Site $site, string $chain_uuid, string $batch_reference_uuid)
    {
        $this->__baseConstruct($site, $chain_uuid, $batch_reference_uuid);
        $this->site = $site; $this->chain_uuid = $chain_uuid; $this->batch_reference_uuid = $batch_reference_uuid;
    }

    public function handle()
    {

        $batch_reference_uuid = $this->batch_reference_uuid;

        // 查找一个尚未处理的站点
        $sites_done = \Illuminate\Support\Facades\DB::table('tracked_jobs')->select('trackable_id')->where('batch_reference_uuid', $batch_reference_uuid)->groupBy('trackable_id')->get()->pluck('trackable_id')->toArray();
        $server = Server::find($this->site->server_id);
        $site_left = $server->sites->whereNotIn('id', $sites_done)->first();

        if($site_left){
            $site = $site_left;
            $chain_uuid = (string) Str::orderedUuid();
            $jobChain = [
                new StartTestJob($site, $chain_uuid, $batch_reference_uuid),
            ];
            $this->batch()->fresh()->add($jobChain);
            // 即使改用dispatch()也无效：
            //$this->batch()->fresh()->add($jobChain)->dispatch();
        }
    }

    public function failed($exception){}
}

内容的提问来源于stack exchange，提问作者urka_mazurka