Question

以下代码遍历整个公司数据库并缓存较低级别的嵌套关系，以便以后可以访问它们。当数据超过约20,000公司和约500,000个嵌套关系时，这会占用大量内存并死亡。如何修改它（将其更改为原始SQL或Laravel查询构建器语句），以便它仅加载我需要的信息并提高其性能？（足以计算条目状态）。

嵌套关系如下

公司＆gt; （有很多）地点＆gt; （有很多）设备＆gt; （具有许多）项目＆gt; （有一个）订单
公司＆gt; （有很多）协议＆gt; （有很多）设备
公司＆gt; （有很多）潜在客户
公司＆gt; （有很多）发票

脚本应该为每个公司输出以下内容的计数

maintainCount - 活动设备的数量（即具有状态 '有效'）对有效协议（即状态为 '未签名'，'有效'或'到期时取消'）
equipmentCount - 活动设备的数量（即带有状态为“活跃”）
leadCount - 活跃线索的数量（即状态为 '打开'）
invoiceCount - 有效发票的数量（即状态为 '待定'或'完成'）
projectCount - 活动项目的数量（即状态为 '待定'或'完成'）
orderCount - 活动订单的数量（即状态为 '待定'或'完成'）

当前脚本如下;

Company::chunk(50, function($companies) {
  foreach ($companies as $company) {
    $maintainedCount = 0;
    $equipmentCount = 0;
    $leadCount = 0;
    $invoiceCount = 0;
    $projectCount = 0;
    $orderCount = 0;
    foreach ($company->agreements as $agreement) {
      if (in_array($agreement->status, ['Unsigned', 'Active', 'Cancel on Expiry'])) {
        $maintainedCount += $agreement->equipment->where('status', 'Active')->count();
      }
    }
    $equipmentCount = $company->equipment->where('status', 'Active')->count();
    $unmaintainedCount = ($equipmentCount - $maintainedCount);
    $leadCount= $company->leads->where('status', 'Open')->count();
    foreach ($company->invoices as $invoice) {
      if (in_array($invoice->status, ['Pending', 'Complete'])) {
        $invoiceCount += 1;
      }
    }    
    foreach ($company->equipment as $equipment) {
      foreach ($equipment->projects as $project) {
        if (in_array($project->status, ['Pending', 'Complete'])) {
          $projectCount += 1;
        }
        if($project->order <> null) {
          if (in_array($project->order->status, ['Pending', 'Complete'])) {
            $orderCount += 1;
          }
        }
      }
    }
    DB::table('companies')->where('id', $company->id)->update(['unmaintained_count' => $unmaintainedCount, 'lead_count' => $leadCount, 'project_count' => $projectCount, 'order_count' => $orderCount, 'invoice_count' => $invoiceCount]);
  }
});
$this->info('Company count cache updated!');

Answer 1

好的，我通过将其转换为直接SQL查询（4gb +到100mb）解决了内存使用问题

DB::connection()->disableQueryLog();
$start = microtime(true);
Company::chunk(50, function($companies) {
    foreach ($companies as $company) {
        $maintainedCount = 0;
        $equipmentCount = 0; 
        $leadCount = 0;
        $invoiceCount = 0;
        $projectCount = 0; 
        $orderCount = 0;
        $maintainedCount = reset(DB::select(DB::raw('SELECT COUNT(*) FROM (
                                    SELECT equipment.id FROM equipment
                                    INNER JOIN agreements ON agreements.id = equipment.agreement_id 
                                    WHERE agreements.company_id = :companyId
                                    AND equipment.status = "Active"
                                    AND (agreements.status = "Active" OR agreements.status = "Unsigned" OR agreements.status = "Cancel on Expiry")) AS QUERY;'), 
                                    ['companyId' => $company->id])[0]);
        $equipmentCount = $company->equipment->where('status', 'Active')->count();
        $unmaintainedCount = ($equipmentCount - $maintainedCount);
        $leadCount= $company->leads->where('status', 'Open')->count();
        $invoiceCount = reset(DB::select(DB::raw('SELECT COUNT(*) FROM (
                                            SELECT invoices.id
                                            FROM invoices
                                            WHERE invoices.company_id = :companyId
                                            AND (invoices.status = "Pending" OR invoices.status = "Complete")) AS QUERY;'), 
                                            ['companyId' => $company->id])[0]);         
        $projectCount = reset(DB::select(DB::raw('SELECT COUNT(*) FROM (
                                                  SELECT projects.id FROM projects
                                                  INNER JOIN equipment ON equipment.id = projects.equipment_id
                                                  INNER JOIN locations ON locations.id = equipment.location_id
                                                  WHERE locations.company_id = :companyId
                                                  AND (projects.status = "Pending" OR projects.status = "Complete")) AS QUERY;'), 
                                                  ['companyId' => $company->id])[0]);               
        $orderCount = reset(DB::select(DB::raw('SELECT COUNT(*) FROM (
                                                SELECT orders.id FROM orders
                                                INNER JOIN projects ON projects.id = orders.project_id
                                                INNER JOIN equipment ON equipment.id = projects.equipment_id
                                                INNER JOIN locations ON locations.id = equipment.location_id
                                                WHERE locations.company_id = :companyId
                                                AND (orders.status = "Pending" OR orders.status = "Complete")) AS QUERY;'), 
                                                ['companyId' => $company->id])[0]);
        DB::table('companies')->where('id', $company->id)->update(['unmaintained_count' => $unmaintainedCount, 'lead_count' => $leadCount, 'project_count' => $projectCount, 'order_count' => $orderCount, 'invoice_count' => $invoiceCount]);
    }
});
$duration = number_format((microtime(true) - $start), 2);
$this->info('Company count cache updated! (it took '.$duration.' seconds)');

如何限制从雄辩查询中检索到的信息并阻止它崩溃？

1 个答案: