Double check for maximum number of workers

[friendica.git] / include / poller.php

diff --git a/include/poller.php b/include/poller.php
index c919b9d2abffff4ef069f5e8d3bc439c461ac004..3b348531c5b6ed47806b1fa821189bb7d0262b17 100644 (file)
--- a/include/poller.php
+++ b/include/poller.php
@@ -26,10 +26,11 @@ function poller_run(&$argv, &$argc){
                unset($db_host, $db_user, $db_pass, $db_data);
        };
 
-       $maxsysload = intval(get_config('system','maxloadavg'));
-       if($maxsysload < 1)
-               $maxsysload = 50;
        if(function_exists('sys_getloadavg')) {
+               $maxsysload = intval(get_config('system','maxloadavg'));
+               if($maxsysload < 1)
+                       $maxsysload = 50;
+
                $load = sys_getloadavg();
                if(intval($load[0]) > $maxsysload) {
                        logger('system: load ' . $load[0] . ' too high. poller deferred to next scheduled run.');
@@ -50,37 +51,42 @@ function poller_run(&$argv, &$argc){
                        if (!posix_kill($pid["pid"], 0))
                                q("UPDATE `workerqueue` SET `executed` = '0000-00-00 00:00:00', `pid` = 0 WHERE `pid` = %d",
                                        intval($pid["pid"]));
+                       else {
+                               // To-Do: Kill long running processes
+                               // But: Update processes (like the database update) mustn't be killed
+                       }
+
+       } else {
+               // Checking the number of workers
+               if (poller_too_much_workers(1))
+                       return;
 
-       } else
-               // Sleep two seconds before checking for running processes to avoid having too many workers
+               // Sleep four seconds before checking for running processes again to avoid having too many workers
                sleep(4);
+       }
 
        // Checking number of workers
-       $workers = q("SELECT COUNT(*) AS `workers` FROM `workerqueue` WHERE `executed` != '0000-00-00 00:00:00'");
-
-       $queues = intval(get_config("system", "worker_queues"));
-
-       if ($queues == 0)
-               $queues = 4;
-
-       if ($workers[0]["workers"] >= $queues)
+       if (poller_too_much_workers(2))
                return;
 
-       while ($r = q("SELECT * FROM `workerqueue` WHERE `executed` = '0000-00-00 00:00:00' ORDER BY `created` LIMIT 1")) {
+       $starttime = time();
 
-               if(function_exists('sys_getloadavg')) {
-                       $load = sys_getloadavg();
-                       if(intval($load[0]) > $maxsysload) {
-                               logger('system: load ' . $load[0] . ' too high. poller deferred to next scheduled run.');
-                               return;
-                       }
-               }
+       while ($r = q("SELECT * FROM `workerqueue` WHERE `executed` = '0000-00-00 00:00:00' ORDER BY `created` LIMIT 1")) {
 
-               q("UPDATE `workerqueue` SET `executed` = '%s', `pid` = %d WHERE `id` = %d",
+               q("UPDATE `workerqueue` SET `executed` = '%s', `pid` = %d WHERE `id` = %d AND `executed` = '0000-00-00 00:00:00'",
                        dbesc(datetime_convert()),
                        intval(getmypid()),
                        intval($r[0]["id"]));
 
+               // Assure that there are no tasks executed twice
+               $id = q("SELECT `id` FROM `workerqueue` WHERE `id` = %d AND `pid` = %d",
+                       intval($r[0]["id"]),
+                       intval(getmypid()));
+               if (!$id) {
+                       logger("Queue item ".$r[0]["id"]." was executed multiple times - skip this execution", LOGGER_DEBUG);
+                       continue;
+               }
+
                $argv = json_decode($r[0]["parameter"]);
 
                $argc = count($argv);
@@ -107,10 +113,55 @@ function poller_run(&$argv, &$argc){
                        q("DELETE FROM `workerqueue` WHERE `id` = %d", intval($r[0]["id"]));
                } else
                        logger("Function ".$funcname." does not exist");
+
+               // Quit the poller once every hour
+               if (time() > ($starttime + 3600))
+                       return;
+
+               // Count active workers and compare them with a maximum value that depends on the load
+               if (poller_too_much_workers(3))
+                       return;
        }
 
 }
 
+function poller_too_much_workers($stage) {
+
+       $queues = get_config("system", "worker_queues");
+
+       if ($queues == 0)
+               $queues = 4;
+
+       $active = poller_active_workers();
+
+       // Decrease the number of workers at higher load
+       if(function_exists('sys_getloadavg')) {
+               $load = max(sys_getloadavg());
+
+               $maxsysload = intval(get_config('system','maxloadavg'));
+               if($maxsysload < 1)
+                       $maxsysload = 50;
+
+               $maxworkers = $queues;
+
+               // Some magical mathemathics to reduce the workers
+               $exponent = 3;
+               $slope = $maxworkers / pow($maxsysload, $exponent);
+               $queues = ceil($slope * pow(max(0, $maxsysload - $load), $exponent));
+
+               logger("Current load stage ".$stage.": ".$load." - maximum: ".$maxsysload." - current queues: ".$active." - maximum: ".$queues, LOGGER_DEBUG);
+
+       }
+
+       return($active >= $queues);
+}
+
+function poller_active_workers() {
+       $workers = q("SELECT COUNT(*) AS `workers` FROM `workerqueue` WHERE `executed` != '0000-00-00 00:00:00'");
+
+       return($workers[0]["workers"]);
+}
+
 if (array_search(__file__,get_included_files())===0){
   poller_run($_SERVER["argv"],$_SERVER["argc"]);
   killme();