(45937) capped list

CronSync.php.example

@@ -25,6 +25,7 @@ Settings::configure([
     'api_key' => 'put_your_key_here',
     'db_path' => '/tmp/anticrawler.sqlite',
     'visitor_forget_after' => 60 * 60 * 24 * 30,
+    // 'pending_requests_max_length' => 200000,
     // 'requests_backend' => 'keydb',
     // 'keydb_host' => '127.0.0.1',
     // 'keydb_port' => 6379,

KeyDBManager.php

@@ -50,7 +50,9 @@ public static function storeRequest(RequestDto $request, ResultDto $result): voi
             'request_status' => $result->status,
         ], JSON_UNESCAPED_SLASHES);
 
-        self::connectFromSettings()->rPush(self::pendingKey(), $payload === false ? '{}' : $payload);
+        $redis = self::connectFromSettings();
+        $redis->rPush(self::pendingKey(), $payload === false ? '{}' : $payload);
+        $redis->lTrim(self::pendingKey(), -self::pendingRequestsMaxLength(), -1);
     }
 
     public static function countPendingRequests(): int
@@ -218,4 +220,9 @@ private static function visitorTtl(): int
     {
         return max(1, (int)Settings::$visitorForgetAfter);
     }
+
+    private static function pendingRequestsMaxLength(): int
+    {
+        return max(1, (int)Settings::$pendingRequestsMaxLength);
+    }
 }

README.md

@@ -67,6 +67,7 @@ List of settings:
 | min_sync_interval | int | Minimum time interval between synchronizations when using default sync behavior, in seconds |
 | max_sync_interval | int | Maximum time interval between synchronizations when using default sync behavior, in seconds |
 | visitor_forget_after | int | Time limit for storing visitor data in the library database, in seconds (decrease this if you have storage issues) |
+| pending_requests_max_length | int | Maximum number of request entries retained in the KeyDB pending queue. Defaults to `100000`; older entries are trimmed away |
 | max_rows_before_sync | int | Maximum number of requests stored between synchronizations when using default sync behavior |
 | sync_by_cron | bool | Set this to true to use the cron synchronization mechanism. See `CronSync.php.example` |
 | requests_backend | string | Backend switch for both request logs and visitor-state data: `sqlite` (default) or `keydb` |
@@ -93,5 +94,5 @@ SQLite can generate high I/O load when the traffic is high. If you see this, you
     ]);
 ```
 
-In this mode, request logs and visitor presence data are stored in KeyDB. `visitor_forget_after` is applied as visitor key TTL.
+In this mode, request logs and visitor presence data are stored in KeyDB. `visitor_forget_after` is applied as visitor key TTL, and `pending_requests_max_length` caps the pending request list by count.
 SQLite is still used for `kv`, `lists`, and `user_agents` tables.

Settings.php

@@ -22,6 +22,9 @@ class Settings
     /** @var int */
     public static $visitorForgetAfter = 60 * 60 * 24 * 30; // One month
 
+    /** @var int */
+    public static $pendingRequestsMaxLength = 100000;
+
     /** @var int */
     public static $maxRowsBeforeSync = 20000;
 
@@ -69,6 +72,8 @@ public static function configure(array $options): void
             self::$minSyncInterval    = $options['min_sync_interval']    ?? self::$minSyncInterval;
             self::$maxSyncInterval    = $options['max_sync_interval']    ?? self::$maxSyncInterval;
             self::$visitorForgetAfter = $options['visitor_forget_after'] ?? self::$visitorForgetAfter;
+            self::$pendingRequestsMaxLength = $options['pending_requests_max_length']
+                ?? self::$pendingRequestsMaxLength;
             self::$maxRowsBeforeSync  = $options['max_rows_before_sync'] ?? self::$maxRowsBeforeSync;
             self::$syncByCron         = $options['sync_by_cron']         ?? self::$syncByCron;
             self::$requestsBackend    = $options['requests_backend']     ?? self::$requestsBackend;