X-Git-Url: https://git.mxchange.org/?a=blobdiff_plain;f=include%2Fpoller.php;h=d134f30d0fac270001d5cac8d9e98bc1411d81d7;hb=8a11a60932115725d65791d99a817aefbde174b7;hp=0edda8170ae04a1ac205ade2f3d89ddfdfead28f;hpb=d87323c2bb54cd7e9b17a15e9926609ee07dc345;p=friendica.git

diff --git a/include/poller.php b/include/poller.php
index 0edda8170a..d134f30d0f 100644
--- a/include/poller.php
+++ b/include/poller.php
@@ -1,7 +1,16 @@
 <?php
+if (!file_exists("boot.php") AND (sizeof($_SERVER["argv"]) != 0)) {
+	$directory = dirname($_SERVER["argv"][0]);
 
-require_once("boot.php");
+	if (substr($directory, 0, 1) != "/")
+		$directory = $_SERVER["PWD"]."/".$directory;
+
+	$directory = realpath($directory."/..");
+
+	chdir($directory);
+}
 
+require_once("boot.php");
 
 function poller_run(&$argv, &$argc){
 	global $a, $db;
@@ -11,296 +20,258 @@ function poller_run(&$argv, &$argc){
 	}
 
 	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);
-  	};
-
-
-	require_once('include/session.php');
-	require_once('include/datetime.php');
-	require_once('library/simplepie/simplepie.inc');
-	require_once('include/items.php');
-	require_once('include/Contact.php');
-	require_once('include/email.php');
-	require_once('include/socgraph.php');
-	require_once('include/pidfile.php');
-
-	load_config('config');
-	load_config('system');
-
-	$maxsysload = intval(get_config('system','maxloadavg'));
-	if($maxsysload < 1)
-		$maxsysload = 50;
-	if(function_exists('sys_getloadavg')) {
-		$load = sys_getloadavg();
-		if(intval($load[0]) > $maxsysload) {
-			logger('system: load ' . $load . ' too high. Poller deferred to next scheduled run.');
-			return;
-		}
-	}
+		@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);
+	};
 
-	$lockpath = get_lockpath();
-	if ($lockpath != '') {
-		$pidfile = new pidfile($lockpath, 'poller');
-		if($pidfile->is_already_running()) {
-			logger("poller: Already running");
-			if ($pidfile->running_time() > 9*60) {
-                                $pidfile->kill();
-                                logger("poller: killed stale process");
-                                // Calling a new instance
-                                proc_run('php','include/poller.php');
-                        }
-			exit;
-		}
-	}
+	if ($a->max_processes_reached())
+		return;
 
+	if (poller_max_connections_reached())
+		return;
 
+	if (App::maxload_reached())
+		return;
 
-	$a->set_baseurl(get_config('system','url'));
+	// Checking the number of workers
+	if (poller_too_much_workers(1)) {
+		poller_kill_stale_workers();
+		return;
+	}
 
-	load_hooks();
+	if(($argc <= 1) OR ($argv[1] != "no_cron")) {
+		// Run the cron job that calls all other jobs
+		proc_run("php","include/cron.php");
 
-	logger('poller: start');
+		// Run the cronhooks job separately from cron for being able to use a different timing
+		proc_run("php","include/cronhooks.php");
 
-	// run queue delivery process in the background
+		// Cleaning dead processes
+		poller_kill_stale_workers();
+	} else
+		// Sleep four seconds before checking for running processes again to avoid having too many workers
+		sleep(4);
 
-	proc_run('php',"include/queue.php");
+	// Checking number of workers
+	if (poller_too_much_workers(2))
+		return;
 
-	// run diaspora photo queue process in the background
+	$starttime = time();
 
-	proc_run('php',"include/dsprphotoq.php");
+	while ($r = q("SELECT * FROM `workerqueue` WHERE `executed` = '0000-00-00 00:00:00' ORDER BY `created` LIMIT 1")) {
 
-	// expire any expired accounts
+		// Constantly check the number of parallel database processes
+		if ($a->max_processes_reached())
+			return;
 
-	q("UPDATE user SET `account_expired` = 1 where `account_expired` = 0 
-		AND `account_expires_on` != '0000-00-00 00:00:00' 
-		AND `account_expires_on` < UTC_TIMESTAMP() ");
+		// Constantly check the number of available database connections to let the frontend be accessible at any time
+		if (poller_max_connections_reached())
+			return;
 
-	// delete user and contact records for recently removed accounts
+		// Count active workers and compare them with a maximum value that depends on the load
+		if (poller_too_much_workers(3))
+			return;
 
-	$r = q("SELECT * FROM `user` WHERE `account_removed` = 1 AND `account_expires_on` < UTC_TIMESTAMP() - INTERVAL 3 DAY");
-	if ($r) {
-		foreach($r as $user) {
-			q("DELETE FROM `contact` WHERE `uid` = %d", intval($user['uid']));
-			q("DELETE FROM `user` WHERE `uid` = %d", intval($user['uid']));
+		q("UPDATE `workerqueue` SET `executed` = '%s', `pid` = %d WHERE `id` = %d AND `executed` = '0000-00-00 00:00:00'",
+			dbesc(datetime_convert()),
+			intval(getmypid()),
+			intval($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;
 		}
-	}
 
-	$abandon_days = intval(get_config('system','account_abandon_days'));
-	if($abandon_days < 1)
-		$abandon_days = 0;
+		$argv = json_decode($r[0]["parameter"]);
 
-	// Check OStatus conversations
-	check_conversations();
+		$argc = count($argv);
 
-	// To-Do: Regenerate usage statistics
-	// q("ANALYZE TABLE `item`");
+		// Check for existance and validity of the include file
+		$include = $argv[0];
 
-	// once daily run birthday_updates and then expire in background
+		if (!validate_include($include)) {
+			logger("Include file ".$argv[0]." is not valid!");
+			q("DELETE FROM `workerqueue` WHERE `id` = %d", intval($r[0]["id"]));
+			continue;
+		}
 
-	$d1 = get_config('system','last_expire_day');
-	$d2 = intval(datetime_convert('UTC','UTC','now','d'));
+		require_once($include);
 
-	if($d2 != intval($d1)) {
+		$funcname=str_replace(".php", "", basename($argv[0]))."_run";
 
-		update_contact_birthdays();
+		if (function_exists($funcname)) {
+			logger("Process ".getmypid()." - ID ".$r[0]["id"].": ".$funcname." ".$r[0]["parameter"]);
+			$funcname($argv, $argc);
 
-		update_suggestions();
+			logger("Process ".getmypid()." - ID ".$r[0]["id"].": ".$funcname." - done");
 
-		set_config('system','last_expire_day',$d2);
-		proc_run('php','include/expire.php');
-	}
+			q("DELETE FROM `workerqueue` WHERE `id` = %d", intval($r[0]["id"]));
+		} else
+			logger("Function ".$funcname." does not exist");
 
-	$last = get_config('system','cache_last_cleared');
-
- 	if($last) {
-		$next = $last + (3600); // Once per hour
-		$clear_cache = ($next <= time());
-        } else
-		$clear_cache = true;
+		// Quit the poller once every hour
+		if (time() > ($starttime + 3600))
+			return;
+	}
 
-	if ($clear_cache) {
-		// clear old cache
-		Cache::clear();
+}
 
-		// clear old item cache files
-		clear_cache();
+/**
+ * @brief Checks if the number of database connections has reached a critical limit.
+ *
+ * @return bool Are more than 3/4 of the maximum connections used?
+ */
+function poller_max_connections_reached() {
+
+	// Fetch the max value from the config. This is needed when the system cannot detect the correct value by itself.
+	$max = get_config("system", "max_connections");
+
+	// Fetch the percentage level where the poller will get active
+	$maxlevel = get_config("system", "max_connections_level");
+	if ($maxlevel == 0)
+		$maxlevel = 75;
+
+	if ($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)
+			$max = $r[0]["Value"];
+
+		// Or it can be granted. This overrides the system variable
+		$r = q("SHOW GRANTS");
+		if ($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))
+						$max = $match[1];
+			}
+	}
 
-		// clear cache for photos
-		clear_cache($a->get_basepath(), $a->get_basepath()."/photo");
+	// 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) {
+		$r = q("SHOW PROCESSLIST");
+		if (!$r)
+			return false;
 
-		// clear smarty cache
-		clear_cache($a->get_basepath()."/view/smarty3/compiled", $a->get_basepath()."/view/smarty3/compiled");
+		$used = count($r);
 
-		// clear cache for image proxy
-		if (!get_config("system", "proxy_disabled")) {
-			clear_cache($a->get_basepath(), $a->get_basepath()."/proxy");
+		logger("Connection usage (user values): ".$used."/".$max, LOGGER_DEBUG);
 
-			$cachetime = get_config('system','proxy_cache_time');
-			if (!$cachetime) $cachetime = PROXY_DEFAULT_TIME;
+		$level = ($used / $max) * 100;
 
-			q('DELETE FROM `photo` WHERE `uid` = 0 AND `resource-id` LIKE "pic:%%" AND `created` < NOW() - INTERVAL %d SECOND', $cachetime);
+		if ($level >= $maxlevel) {
+			logger("Maximum level (".$maxlevel."%) of user connections reached: ".$used."/".$max);
+			return true;
 		}
-
-		set_config('system','cache_last_cleared', time());
 	}
 
-	$manual_id  = 0;
-	$generation = 0;
-	$hub_update = false;
-	$force      = false;
-	$restart    = false;
+	// 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)
+		return false;
 
-	if(($argc > 1) && ($argv[1] == 'force'))
-		$force = true;
+	$max = intval($r[0]["Value"]);
+	if ($max == 0)
+		return false;
 
-	if(($argc > 1) && ($argv[1] == 'restart')) {
-		$restart = true;
-		$generation = intval($argv[2]);
-		if(! $generation)
-			killme();
-	}
+	$r = q("SHOW STATUS WHERE `variable_name` = 'Threads_connected'");
+	if (!$r)
+		return false;
 
-	if(($argc > 1) && intval($argv[1])) {
-		$manual_id = intval($argv[1]);
-		$force     = true;
-	}
+	$used = intval($r[0]["Value"]);
+	if ($used == 0)
+		return false;
 
-	$interval = intval(get_config('system','poll_interval'));
-	if(! $interval) 
-		$interval = ((get_config('system','delivery_interval') === false) ? 3 : intval(get_config('system','delivery_interval')));
+	logger("Connection usage (system values): ".$used."/".$max, LOGGER_DEBUG);
 
-	$sql_extra = (($manual_id) ? " AND `id` = $manual_id " : "");
+	$level = $used / $max * 100;
 
-	reload_plugins();
+	if ($level < $maxlevel)
+		return false;
 
-	$d = datetime_convert();
+	logger("Maximum level (".$level."%) of system connections reached: ".$used."/".$max);
+	return true;
+}
 
-	if(! $restart)
-		proc_run('php','include/cronhooks.php');
+/**
+ * @brief fix the queue entry if the worker process died
+ *
+ */
+function poller_kill_stale_workers() {
+	$r = q("SELECT `pid`, `executed` FROM `workerqueue` WHERE `executed` != '0000-00-00 00:00:00'");
 
-	// Only poll from those with suitable relationships,
-	// and which have a polling address and ignore Diaspora since 
-	// we are unable to match those posts with a Diaspora GUID and prevent duplicates.
+	if (!dbm::is_result($r)) {
+		// No processing here needed
+		return;
+	}
 
-	$abandon_sql = (($abandon_days) 
-		? sprintf(" AND `user`.`login_date` > UTC_TIMESTAMP() - INTERVAL %d DAY ", intval($abandon_days)) 
-		: '' 
-	);
+	foreach($r AS $pid)
+		if (!posix_kill($pid["pid"], 0))
+			q("UPDATE `workerqueue` SET `executed` = '0000-00-00 00:00:00', `pid` = 0 WHERE `pid` = %d",
+				intval($pid["pid"]));
+		else {
+			// Kill long running processes
+			$duration = (time() - strtotime($pid["executed"])) / 60;
+			if ($duration > 180) {
+				logger("Worker process ".$pid["pid"]." took more than 3 hours. It will be killed now.");
+				posix_kill($pid["pid"], SIGTERM);
+
+				// Question: If a process is stale: Should we remove it or should we reschedule it?
+				// By now we rescheduling it. It's maybe not the wisest decision?
+				q("UPDATE `workerqueue` SET `executed` = '0000-00-00 00:00:00', `pid` = 0 WHERE `pid` = %d",
+					intval($pid["pid"]));
+			} else
+				logger("Worker process ".$pid["pid"]." now runs for ".round($duration)." minutes. That's okay.", LOGGER_DEBUG);
+		}
+}
 
-	$contacts = q("SELECT `contact`.`id` FROM `contact` INNER JOIN `user` ON `user`.`uid` = `contact`.`uid` 
-		WHERE ( `rel` = %d OR `rel` = %d ) AND `poll` != ''
-		AND NOT `network` IN ( '%s', '%s', '%s' )
-		$sql_extra 
-		AND `self` = 0 AND `contact`.`blocked` = 0 AND `contact`.`readonly` = 0 
-		AND `contact`.`archive` = 0 
-		AND `user`.`account_expired` = 0 AND `user`.`account_removed` = 0 $abandon_sql ORDER BY RAND()",
-		intval(CONTACT_IS_SHARING),
-		intval(CONTACT_IS_FRIEND),
-		dbesc(NETWORK_DIASPORA),
-		dbesc(NETWORK_FACEBOOK),
-		dbesc(NETWORK_PUMPIO)
-	);
+function poller_too_much_workers($stage) {
 
-	if(! count($contacts)) {
-		return;
-	}
+	$queues = get_config("system", "worker_queues");
 
-	foreach($contacts as $c) {
+	if ($queues == 0)
+		$queues = 4;
 
-		$res = q("SELECT * FROM `contact` WHERE `id` = %d LIMIT 1",
-			intval($c['id'])
-		);
+	$active = poller_active_workers();
 
-		if((! $res) || (! count($res)))
-			continue;
+	// Decrease the number of workers at higher load
+	$load = current_load();
+	if($load) {
+		$maxsysload = intval(get_config('system','maxloadavg'));
+		if($maxsysload < 1)
+			$maxsysload = 50;
 
-		foreach($res as $contact) {
-
-			$xml = false;
-
-			if($manual_id)
-				$contact['last-update'] = '0000-00-00 00:00:00';
-
-			if($contact['network'] === NETWORK_DFRN)
-				$contact['priority'] = 2;
-
-			if(!get_config('system','ostatus_use_priority') and ($contact['network'] === NETWORK_OSTATUS))
-				$contact['priority'] = 2;
-
-			if($contact['priority'] || $contact['subhub']) {
-
-				$hub_update = true;
-				$update     = false;
-
-				$t = $contact['last-update'];
-
-				// We should be getting everything via a hub. But just to be sure, let's check once a day.
-				// (You can make this more or less frequent if desired by setting 'pushpoll_frequency' appropriately)
-				// This also lets us update our subscription to the hub, and add or replace hubs in case it
-				// changed. We will only update hubs once a day, regardless of 'pushpoll_frequency'.
-
-
-				if($contact['subhub']) {
-					$poll_interval = get_config('system','pushpoll_frequency');
-					$contact['priority'] = (($poll_interval !== false) ? intval($poll_interval) : 3);
-					$hub_update = false;
-
-					if((datetime_convert('UTC','UTC', 'now') > datetime_convert('UTC','UTC', $t . " + 1 day")) || $force)
-							$hub_update = true;
-				}
-				else
-					$hub_update = false;
-
-				/**
-				 * Based on $contact['priority'], should we poll this site now? Or later?
-				 */
-
-				switch ($contact['priority']) {
-					case 5:
-						if(datetime_convert('UTC','UTC', 'now') > datetime_convert('UTC','UTC', $t . " + 1 month"))
-							$update = true;
-						break;
-					case 4:
-						if(datetime_convert('UTC','UTC', 'now') > datetime_convert('UTC','UTC', $t . " + 1 week"))
-							$update = true;
-						break;
-					case 3:
-						if(datetime_convert('UTC','UTC', 'now') > datetime_convert('UTC','UTC', $t . " + 1 day"))
-							$update = true;
-						break;
-					case 2:
-						if(datetime_convert('UTC','UTC', 'now') > datetime_convert('UTC','UTC', $t . " + 12 hour"))
-							$update = true;
-						break;
-					case 1:
-					default:
-						if(datetime_convert('UTC','UTC', 'now') > datetime_convert('UTC','UTC', $t . " + 1 hour"))
-							$update = true;
-						break;
-				}
-				if((! $update) && (! $force))
-					continue;
-			}
+		$maxworkers = $queues;
 
-			// Don't run onepoll.php if the contact isn't pollable
-			// This check also is inside the onepoll.php - but this will reduce the load
-			if (in_array($contact["rel"], array(CONTACT_IS_SHARING, CONTACT_IS_FRIEND)) AND ($contact["poll"] != "")
-				AND !in_array($contact['network'], array(NETWORK_DIASPORA, NETWORK_FACEBOOK, NETWORK_PUMPIO, NETWORK_TWITTER, NETWORK_APPNET))
-				AND !$contact["self"] AND !$contact["blocked"] AND !$contact["readonly"] AND !$contact["archive"])
-				proc_run('php','include/onepoll.php',$contact['id']);
+		// 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);
 
-			if($interval)
-				@time_sleep_until(microtime(true) + (float) $interval);
-		}
 	}
 
-	return;
+	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($argv,$argc);
+  poller_run($_SERVER["argv"],$_SERVER["argc"]);
   killme();
 }
+?>