]> git.mxchange.org Git - friendica.git/blobdiff - include/poller.php
Fetching new queue tasks in a bulk to increase speed
[friendica.git] / include / poller.php
index 0600434a715dd17cbdd0ec2c1d73c1c94e58ce63..4801ea8fdf09aaa20d1998b63b0d4162ccf18b0e 100644 (file)
@@ -2,13 +2,14 @@
 
 use Friendica\App;
 use Friendica\Core\Config;
+use Friendica\Util\Lock;
 
-if (!file_exists("boot.php") AND (sizeof($_SERVER["argv"]) != 0)) {
+if (!file_exists("boot.php") && (sizeof($_SERVER["argv"]) != 0)) {
        $directory = dirname($_SERVER["argv"][0]);
 
-       if (substr($directory, 0, 1) != "/")
+       if (substr($directory, 0, 1) != "/") {
                $directory = $_SERVER["PWD"]."/".$directory;
-
+       }
        $directory = realpath($directory."/..");
 
        chdir($directory);
@@ -19,16 +20,12 @@ require_once("boot.php");
 function poller_run($argv, $argc){
        global $a, $db;
 
-       if (is_null($a)) {
-               $a = new App(dirname(__DIR__));
-       }
+       $a = new App(dirname(__DIR__));
 
-       if (is_null($db)) {
-               @include(".htconfig.php");
-               require_once("include/dba.php");
-               $db = new dba($db_host, $db_user, $db_pass, $db_data);
-               unset($db_host, $db_user, $db_pass, $db_data);
-       };
+       @include(".htconfig.php");
+       require_once("include/dba.php");
+       $db = new dba($db_host, $db_user, $db_pass, $db_data);
+       unset($db_host, $db_user, $db_pass, $db_data);
 
        Config::load();
 
@@ -41,56 +38,122 @@ function poller_run($argv, $argc){
 
        load_hooks();
 
+       // At first check the maximum load. We shouldn't continue with a high load
+       if ($a->maxload_reached()) {
+               logger('Pre check: maximum load reached, quitting.', LOGGER_DEBUG);
+               return;
+       }
+
+       // We now start the process. This is done after the load check since this could increase the load.
        $a->start_process();
 
-       if ($a->min_memory_reached()) {
-               return;
+       // Kill stale processes every 5 minutes
+       $last_cleanup = Config::get('system', 'poller_last_cleaned', 0);
+       if (time() > ($last_cleanup + 300)) {
+               Config::set('system', 'poller_last_cleaned', time());
+               poller_kill_stale_workers();
        }
 
-       if (poller_max_connections_reached()) {
+       // Count active workers and compare them with a maximum value that depends on the load
+       if (poller_too_much_workers()) {
+               logger('Pre check: Active worker limit reached, quitting.', LOGGER_DEBUG);
                return;
        }
 
-       if ($a->maxload_reached()) {
+       // Do we have too few memory?
+       if ($a->min_memory_reached()) {
+               logger('Pre check: Memory limit reached, quitting.', LOGGER_DEBUG);
                return;
        }
 
-       if (($argc <= 1) OR ($argv[1] != "no_cron")) {
-               poller_run_cron();
+       // Possibly there are too much database connections
+       if (poller_max_connections_reached()) {
+               logger('Pre check: maximum connections reached, quitting.', LOGGER_DEBUG);
+               return;
        }
 
+       // Possibly there are too much database processes that block the system
        if ($a->max_processes_reached()) {
+               logger('Pre check: maximum processes reached, quitting.', LOGGER_DEBUG);
                return;
        }
 
-       // Checking the number of workers
-       if (poller_too_much_workers()) {
-               poller_kill_stale_workers();
-               return;
+       // Now we start additional cron processes if we should do so
+       if (($argc <= 1) || ($argv[1] != "no_cron")) {
+               poller_run_cron();
        }
 
        $starttime = time();
 
+       // We fetch the next queue entry that is about to be executed
        while ($r = poller_worker_process()) {
-
-               // Check free memory
-               if ($a->min_memory_reached()) {
-                       return;
+               foreach ($r AS $entry) {
+                       // Assure that the priority is an integer value
+                       $entry['priority'] = (int)$entry['priority'];
+
+                       // The work will be done
+                       if (!poller_execute($entry)) {
+                               logger('Process execution failed, quitting.', LOGGER_DEBUG);
+                               return;
+                       }
                }
 
-               // Count active workers and compare them with a maximum value that depends on the load
-               if (poller_too_much_workers()) {
-                       return;
-               }
+               // To avoid the quitting of multiple pollers only one poller at a time will execute the check
+               if (Lock::set('poller_worker', 0)) {
+                       // Count active workers and compare them with a maximum value that depends on the load
+                       if (poller_too_much_workers()) {
+                               logger('Active worker limit reached, quitting.', LOGGER_DEBUG);
+                               return;
+                       }
 
-               if (!poller_execute($r[0])) {
-                       return;
+                       // Check free memory
+                       if ($a->min_memory_reached()) {
+                               logger('Memory limit reached, quitting.', LOGGER_DEBUG);
+                               return;
+                       }
+                       Lock::remove('poller_worker');
                }
 
-               // Quit the poller once every hour
-               if (time() > ($starttime + 3600))
+               // Quit the poller once every 5 minutes
+               if (time() > ($starttime + 300)) {
+                       logger('Process lifetime reached, quitting.', LOGGER_DEBUG);
                        return;
+               }
        }
+       logger("Couldn't select a workerqueue entry, quitting.", LOGGER_DEBUG);
+}
+
+/**
+ * @brief Returns the number of non executed entries in the worker queue
+ *
+ * @return integer Number of non executed entries in the worker queue
+ */
+function poller_total_entries() {
+       $s = q("SELECT COUNT(*) AS `total` FROM `workerqueue` WHERE `executed` <= '%s'", dbesc(NULL_DATE));
+       return $s[0]["total"];
+}
+
+/**
+ * @brief Returns the highest priority in the worker queue that isn't executed
+ *
+ * @return integer Number of active poller processes
+ */
+function poller_highest_priority() {
+       $s = q("SELECT `priority` FROM `workerqueue` WHERE `executed` <= '%s' ORDER BY `priority` LIMIT 1", dbesc(NULL_DATE));
+       return $s[0]["priority"];
+}
+
+/**
+ * @brief Returns if a process with the given priority is running
+ *
+ * @param integer $priority The priority that should be checked
+ *
+ * @return integer Is there a process running with that priority?
+ */
+function poller_process_with_priority_active($priority) {
+       $s = q("SELECT `id` FROM `workerqueue` WHERE `priority` <= %d AND `executed` > '%s' LIMIT 1",
+                       intval($priority), dbesc(NULL_DATE));
+       return dbm::is_result($s);
 }
 
 /**
@@ -108,43 +171,22 @@ function poller_execute($queue) {
 
        // Quit when in maintenance
        if (Config::get('system', 'maintenance', true)) {
+               logger("Maintenance mode - quit process ".$mypid, LOGGER_DEBUG);
                return false;
        }
 
        // Constantly check the number of parallel database processes
        if ($a->max_processes_reached()) {
+               logger("Max processes reached for process ".$mypid, LOGGER_DEBUG);
                return false;
        }
 
        // Constantly check the number of available database connections to let the frontend be accessible at any time
        if (poller_max_connections_reached()) {
+               logger("Max connection reached for process ".$mypid, LOGGER_DEBUG);
                return false;
        }
 
-       if (!dba::update('workerqueue', array('executed' => datetime_convert(), 'pid' => $mypid),
-                       array('id' => $queue["id"], 'pid' => 0))) {
-               logger("Couldn't update queue entry ".$queue["id"]." - skip this execution", LOGGER_DEBUG);
-               dba::commit();
-               return true;
-       }
-
-       // Assure that there are no tasks executed twice
-       $id = q("SELECT `pid`, `executed` FROM `workerqueue` WHERE `id` = %d", intval($queue["id"]));
-       if (!$id) {
-               logger("Queue item ".$queue["id"]." vanished - skip this execution", LOGGER_DEBUG);
-               dba::commit();
-               return true;
-       } elseif ((strtotime($id[0]["executed"]) <= 0) OR ($id[0]["pid"] == 0)) {
-               logger("Entry for queue item ".$queue["id"]." wasn't stored - skip this execution", LOGGER_DEBUG);
-               dba::commit();
-               return true;
-       } elseif ($id[0]["pid"] != $mypid) {
-               logger("Queue item ".$queue["id"]." is to be executed by process ".$id[0]["pid"]." and not by me (".$mypid.") - skip this execution", LOGGER_DEBUG);
-               dba::commit();
-               return true;
-       }
-       dba::commit();
-
        $argv = json_decode($queue["parameter"]);
 
        // Check for existance and validity of the include file
@@ -161,9 +203,7 @@ function poller_execute($queue) {
        $funcname = str_replace(".php", "", basename($argv[0]))."_run";
 
        if (function_exists($funcname)) {
-
                poller_exec_function($queue, $funcname, $argv);
-
                dba::delete('workerqueue', array('id' => $queue["id"]));
        } else {
                logger("Function ".$funcname." does not exist");
@@ -210,10 +250,12 @@ function poller_exec_function($queue, $funcname, $argv) {
        // But preserve the old one for the worker
        $old_process_id = $a->process_id;
        $a->process_id = uniqid("wrk", true);
+       $a->queue = $queue;
 
        $funcname($argv, $argc);
 
        $a->process_id = $old_process_id;
+       unset($a->queue);
 
        $duration = number_format(microtime(true) - $stamp, 3);
 
@@ -238,24 +280,27 @@ function poller_exec_function($queue, $funcname, $argv) {
                                $o = "\nDatabase Read:\n";
                                foreach ($a->callstack["database"] AS $func => $time) {
                                        $time = round($time, 3);
-                                       if ($time > 0)
+                                       if ($time > 0) {
                                                $o .= $func.": ".$time."\n";
+                                       }
                                }
                        }
                        if (isset($a->callstack["database_write"])) {
                                $o .= "\nDatabase Write:\n";
                                foreach ($a->callstack["database_write"] AS $func => $time) {
                                        $time = round($time, 3);
-                                       if ($time > 0)
+                                       if ($time > 0) {
                                                $o .= $func.": ".$time."\n";
+                                       }
                                }
                        }
                        if (isset($a->callstack["network"])) {
                                $o .= "\nNetwork:\n";
                                foreach ($a->callstack["network"] AS $func => $time) {
                                        $time = round($time, 3);
-                                       if ($time > 0)
+                                       if ($time > 0) {
                                                $o .= $func.": ".$time."\n";
+                                       }
                                }
                        }
                } else {
@@ -296,27 +341,30 @@ function poller_max_connections_reached() {
        if ($max == 0) {
                // the maximum number of possible user connections can be a system variable
                $r = q("SHOW VARIABLES WHERE `variable_name` = 'max_user_connections'");
-               if ($r)
+               if (dbm::is_result($r)) {
                        $max = $r[0]["Value"];
-
+               }
                // Or it can be granted. This overrides the system variable
                $r = q("SHOW GRANTS");
-               if ($r)
+               if (dbm::is_result($r)) {
                        foreach ($r AS $grants) {
                                $grant = array_pop($grants);
-                               if (stristr($grant, "GRANT USAGE ON"))
-                                       if (preg_match("/WITH MAX_USER_CONNECTIONS (\d*)/", $grant, $match))
+                               if (stristr($grant, "GRANT USAGE ON")) {
+                                       if (preg_match("/WITH MAX_USER_CONNECTIONS (\d*)/", $grant, $match)) {
                                                $max = $match[1];
+                                       }
+                               }
                        }
+               }
        }
 
        // If $max is set we will use the processlist to determine the current number of connections
        // The processlist only shows entries of the current user
        if ($max != 0) {
                $r = q("SHOW PROCESSLIST");
-               if (!dbm::is_result($r))
+               if (!dbm::is_result($r)) {
                        return false;
-
+               }
                $used = count($r);
 
                logger("Connection usage (user values): ".$used."/".$max, LOGGER_DEBUG);
@@ -332,28 +380,28 @@ function poller_max_connections_reached() {
        // We will now check for the system values.
        // This limit could be reached although the user limits are fine.
        $r = q("SHOW VARIABLES WHERE `variable_name` = 'max_connections'");
-       if (!$r)
+       if (!dbm::is_result($r)) {
                return false;
-
+       }
        $max = intval($r[0]["Value"]);
-       if ($max == 0)
+       if ($max == 0) {
                return false;
-
+       }
        $r = q("SHOW STATUS WHERE `variable_name` = 'Threads_connected'");
-       if (!$r)
+       if (!dbm::is_result($r)) {
                return false;
-
+       }
        $used = intval($r[0]["Value"]);
-       if ($used == 0)
+       if ($used == 0) {
                return false;
-
+       }
        logger("Connection usage (system values): ".$used."/".$max, LOGGER_DEBUG);
 
        $level = $used / $max * 100;
 
-       if ($level < $maxlevel)
+       if ($level < $maxlevel) {
                return false;
-
+       }
        logger("Maximum level (".$level."%) of system connections reached: ".$used."/".$max);
        return true;
 }
@@ -411,6 +459,10 @@ function poller_kill_stale_workers() {
 /**
  * @brief Checks if the number of active workers exceeds the given limits
  *
+ * @param integer $entries The number of not executed entries in the worker queue
+ * @param integer $top_priority The highest not executed priority in the worker queue
+ * @param boolean $high_running Is a process with priority "$top_priority" running?
+ *
  * @return bool Are there too much workers running?
  */
 function poller_too_much_workers() {
@@ -432,48 +484,46 @@ function poller_too_much_workers() {
                $slope = $maxworkers / pow($maxsysload, $exponent);
                $queues = ceil($slope * pow(max(0, $maxsysload - $load), $exponent));
 
-               $s = q("SELECT COUNT(*) AS `total` FROM `workerqueue` WHERE `executed` <= '%s'", dbesc(NULL_DATE));
-               $entries = $s[0]["total"];
+               if (Config::get('system', 'worker_debug')) {
+                       // Create a list of queue entries grouped by their priority
+                       $listitem = array();
+
+                       // Adding all processes with no workerqueue entry
+                       $processes = dba::p("SELECT COUNT(*) AS `running` FROM `process` WHERE NOT EXISTS (SELECT id FROM `workerqueue` WHERE `workerqueue`.`pid` = `process`.`pid`)");
+                       if ($process = dba::fetch($processes)) {
+                               $listitem[0] = "0:".$process["running"];
+                       }
+                       dba::close($processes);
+
+                       // Now adding all processes with workerqueue entries
+                       $entries = dba::p("SELECT COUNT(*) AS `entries`, `priority` FROM `workerqueue` GROUP BY `priority`");
+                       while ($entry = dba::fetch($entries)) {
+                               $processes = dba::p("SELECT COUNT(*) AS `running` FROM `process` INNER JOIN `workerqueue` ON `workerqueue`.`pid` = `process`.`pid` WHERE `priority` = ?", $entry["priority"]);
+                               if ($process = dba::fetch($processes)) {
+                                       $listitem[$entry["priority"]] = $entry["priority"].":".$process["running"]."/".$entry["entries"];
+                               }
+                               dba::close($processes);
+                       }
+                       dba::close($entries);
+                       $processlist = ' ('.implode(', ', $listitem).')';
+               }
 
-               if (Config::get("system", "worker_fastlane", false) AND ($queues > 0) AND ($entries > 0) AND ($active >= $queues)) {
-                       $s = q("SELECT `priority` FROM `workerqueue` WHERE `executed` <= '%s' ORDER BY `priority` LIMIT 1", dbesc(NULL_DATE));
-                       $top_priority = $s[0]["priority"];
+               $entries = poller_total_entries();
 
-                       $s = q("SELECT `id` FROM `workerqueue` WHERE `priority` <= %d AND `executed` > '%s' LIMIT 1",
-                               intval($top_priority), dbesc(NULL_DATE));
-                       $high_running = dbm::is_result($s);
+               if (Config::get("system", "worker_fastlane", false) && ($queues > 0) && ($entries > 0) && ($active >= $queues)) {
+                       $top_priority = poller_highest_priority();
+                       $high_running = poller_process_with_priority_active($top_priority);
 
-                       if (!$high_running AND ($top_priority > PRIORITY_UNDEFINED) AND ($top_priority < PRIORITY_NEGLIGIBLE)) {
+                       if (!$high_running && ($top_priority > PRIORITY_UNDEFINED) && ($top_priority < PRIORITY_NEGLIGIBLE)) {
                                logger("There are jobs with priority ".$top_priority." waiting but none is executed. Open a fastlane.", LOGGER_DEBUG);
                                $queues = $active + 1;
                        }
                }
 
-               // Create a list of queue entries grouped by their priority
-               $running = array(PRIORITY_CRITICAL => 0,
-                               PRIORITY_HIGH => 0,
-                               PRIORITY_MEDIUM => 0,
-                               PRIORITY_LOW => 0,
-                               PRIORITY_NEGLIGIBLE => 0);
-
-               $r = q("SELECT COUNT(*) AS `running`, `priority` FROM `process` INNER JOIN `workerqueue` ON `workerqueue`.`pid` = `process`.`pid` GROUP BY `priority`");
-               if (dbm::is_result($r))
-                       foreach ($r AS $process)
-                               $running[$process["priority"]] = $process["running"];
-
-               $processlist = "";
-               $r = q("SELECT COUNT(*) AS `entries`, `priority` FROM `workerqueue` GROUP BY `priority`");
-               if (dbm::is_result($r))
-                       foreach ($r as $entry) {
-                               if ($processlist != "")
-                                       $processlist .= ", ";
-                               $processlist .= $entry["priority"].":".$running[$entry["priority"]]."/".$entry["entries"];
-                       }
-
-               logger("Load: ".$load."/".$maxsysload." - processes: ".$active."/".$entries." (".$processlist.") - maximum: ".$queues."/".$maxqueues, LOGGER_DEBUG);
+               logger("Load: ".$load."/".$maxsysload." - processes: ".$active."/".$entries.$processlist." - maximum: ".$queues."/".$maxqueues, LOGGER_DEBUG);
 
                // Are there fewer workers running as possible? Then fork a new one.
-               if (!Config::get("system", "worker_dont_fork") AND ($queues > ($active + 1)) AND ($entries > 1)) {
+               if (!Config::get("system", "worker_dont_fork") && ($queues > ($active + 1)) && ($entries > 1)) {
                        logger("Active workers: ".$active."/".$queues." Fork a new worker.", LOGGER_DEBUG);
                        $args = array("include/poller.php", "no_cron");
                        $a = get_app();
@@ -481,7 +531,7 @@ function poller_too_much_workers() {
                }
        }
 
-       return($active >= $queues);
+       return $active >= $queues;
 }
 
 /**
@@ -492,7 +542,7 @@ function poller_too_much_workers() {
 function poller_active_workers() {
        $workers = q("SELECT COUNT(*) AS `processes` FROM `process` WHERE `command` = 'poller.php'");
 
-       return($workers[0]["processes"]);
+       return $workers[0]["processes"];
 }
 
 /**
@@ -513,75 +563,116 @@ function poller_passing_slow(&$highest_priority) {
                INNER JOIN `workerqueue` ON `workerqueue`.`pid` = `process`.`pid`");
 
        // No active processes at all? Fine
-       if (!dbm::is_result($r))
-               return(false);
-
+       if (!dbm::is_result($r)) {
+               return false;
+       }
        $priorities = array();
-       foreach ($r AS $line)
+       foreach ($r AS $line) {
                $priorities[] = $line["priority"];
-
+       }
        // Should not happen
-       if (count($priorities) == 0)
-               return(false);
-
+       if (count($priorities) == 0) {
+               return false;
+       }
        $highest_priority = min($priorities);
 
        // The highest process is already the slowest one?
        // Then we quit
-       if ($highest_priority == PRIORITY_NEGLIGIBLE)
-               return(false);
-
+       if ($highest_priority == PRIORITY_NEGLIGIBLE) {
+               return false;
+       }
        $high = 0;
-       foreach ($priorities AS $priority)
-               if ($priority == $highest_priority)
+       foreach ($priorities AS $priority) {
+               if ($priority == $highest_priority) {
                        ++$high;
-
+               }
+       }
        logger("Highest priority: ".$highest_priority." Total processes: ".count($priorities)." Count high priority processes: ".$high, LOGGER_DEBUG);
        $passing_slow = (($high/count($priorities)) > (2/3));
 
-       if ($passing_slow)
+       if ($passing_slow) {
                logger("Passing slower processes than priority ".$highest_priority, LOGGER_DEBUG);
-
-       return($passing_slow);
+       }
+       return $passing_slow;
 }
 
 /**
- * @brief Returns the next worker process
+ * @brief Find and claim the next worker process for us
  *
- * @return string SQL statement
+ * @return boolean Have we found something?
  */
-function poller_worker_process() {
-
-       dba::transaction();
-
+function find_worker_processes() {
        // Check if we should pass some low priority process
        $highest_priority = 0;
+       $found = false;
 
        if (poller_passing_slow($highest_priority)) {
                // Are there waiting processes with a higher priority than the currently highest?
-               $r = q("SELECT * FROM `workerqueue`
-                               WHERE `executed` <= '%s' AND `priority` < %d
-                               ORDER BY `priority`, `created` LIMIT 1",
-                               dbesc(NULL_DATE),
-                               intval($highest_priority));
-               if (dbm::is_result($r)) {
-                       return $r;
+               $result = dba::p("UPDATE `workerqueue` SET `executed` = ?, `pid` = ?
+                                       WHERE `executed` <= ? AND `priority` < ?
+                                       ORDER BY `priority`, `created` LIMIT 5",
+                               datetime_convert(), getmypid(), NULL_DATE, $highest_priority);
+               if (dbm::is_result($result)) {
+                       $found = (dba::num_rows($result) > 0);
+               }
+               dba::close($result);
+
+               if (!$found) {
+                       // Give slower processes some processing time
+                       $result = dba::p("UPDATE `workerqueue` SET `executed` = ?, `pid` = ?
+                                               WHERE `executed` <= ? AND `priority` > ?
+                                               ORDER BY `priority`, `created` LIMIT 1",
+                                       datetime_convert(), getmypid(), NULL_DATE, $highest_priority);
+                       if (dbm::is_result($result)) {
+                               $found = (dba::num_rows($result) > 0);
+                       }
+                       dba::close($result);
                }
-               // Give slower processes some processing time
-               $r = q("SELECT * FROM `workerqueue`
-                               WHERE `executed` <= '%s' AND `priority` > %d
-                               ORDER BY `priority`, `created` LIMIT 1",
-                               dbesc(NULL_DATE),
-                               intval($highest_priority));
        }
 
        // If there is no result (or we shouldn't pass lower processes) we check without priority limit
-       if (($highest_priority == 0) OR !dbm::is_result($r)) {
-               $r = q("SELECT * FROM `workerqueue` WHERE `executed` <= '%s' ORDER BY `priority`, `created` LIMIT 1", dbesc(NULL_DATE));
+       if (!$found) {
+               $result = dba::p("UPDATE `workerqueue` SET `executed` = ?, `pid` = ? WHERE `executed` <= ? ORDER BY `priority`, `created` LIMIT 5",
+                               datetime_convert(), getmypid(), NULL_DATE);
+               if (dbm::is_result($result)) {
+                       $found = (dba::num_rows($result) > 0);
+               }
+               dba::close($result);
+       }
+       return $found;
+}
+
+/**
+ * @brief Returns the next worker process
+ *
+ * @return string SQL statement
+ */
+function poller_worker_process() {
+
+       $stamp = (float)microtime(true);
+
+       $timeout = 10;
+       do {
+               $found = find_worker_processes();
+       } while (!$found && (poller_total_entries() > 0) && (--$timeout > 0));
+
+       logger('Duration: '.number_format(microtime(true) - $stamp, 3), LOGGER_DEBUG);
+
+       if ($found) {
+               $r = q("SELECT * FROM `workerqueue` WHERE `pid` = %d", intval(getmypid()));
        }
        return $r;
 }
 
+/**
+ * @brief Removes a workerqueue entry from the current process
+ */
+function poller_unclaim_process() {
+       $mypid = getmypid();
+
+       dba::update('workerqueue', array('executed' => NULL_DATE, 'pid' => 0), array('pid' => $mypid));
+}
+
 /**
  * @brief Call the front end worker
  */
@@ -678,7 +769,11 @@ function poller_run_cron() {
 if (array_search(__file__,get_included_files())===0){
        poller_run($_SERVER["argv"],$_SERVER["argc"]);
 
+       poller_unclaim_process();
+
        get_app()->end_process();
 
+       Lock::remove('poller_worker');
+
        killme();
 }