friendica/src/Core/Worker.php

1280 lines
40 KiB
PHP
Raw Normal View History

<?php
/**
* @file src/Core/Worker.php
*/
namespace Friendica\Core;
use Friendica\BaseObject;
use Friendica\Database\DBA;
use Friendica\Model\Process;
use Friendica\Util\DateTimeFormat;
use Friendica\Util\Logger\WorkerLogger;
2018-06-26 17:42:26 -04:00
use Friendica\Util\Network;
/**
2017-11-05 07:15:53 -05:00
* @file src/Core/Worker.php
*
2017-11-06 10:32:34 -05:00
* @brief Contains the class for the worker background job processing
*/
/**
* @brief Worker methods
*/
class Worker
{
private static $up_start;
2019-02-09 18:10:15 -05:00
private static $db_duration = 0;
private static $db_duration_count = 0;
private static $db_duration_write = 0;
private static $db_duration_stat = 0;
private static $lock_duration = 0;
private static $last_update;
private static $mode = 0;
/**
* @brief Processes the tasks that are in the workerqueue table
*
* @param boolean $run_cron Should the cron processes be executed?
* @return void
2019-01-06 16:06:53 -05:00
* @throws \Friendica\Network\HTTPException\InternalServerErrorException
*/
public static function processQueue($run_cron = true)
{
$a = \get_app();
// Ensure that all "strtotime" operations do run timezone independent
date_default_timezone_set('UTC');
self::$up_start = microtime(true);
2017-11-05 10:28:55 -05:00
// At first check the maximum load. We shouldn't continue with a high load
if ($a->isMaxLoadReached()) {
Logger::log('Pre check: maximum load reached, quitting.', Logger::DEBUG);
2017-11-05 10:28:55 -05:00
return;
}
// We now start the process. This is done after the load check since this could increase the load.
self::startProcess();
2017-11-05 10:28:55 -05:00
// Kill stale processes every 5 minutes
$last_cleanup = Config::get('system', 'worker_last_cleaned', 0);
if (time() > ($last_cleanup + 300)) {
Config::set('system', 'worker_last_cleaned', time());
self::killStaleWorkers();
}
// Count active workers and compare them with a maximum value that depends on the load
if (self::tooMuchWorkers()) {
Logger::log('Pre check: Active worker limit reached, quitting.', Logger::DEBUG);
return;
}
// Do we have too few memory?
if ($a->isMinMemoryReached()) {
Logger::log('Pre check: Memory limit reached, quitting.', Logger::DEBUG);
return;
}
// Possibly there are too much database connections
if (self::maxConnectionsReached()) {
Logger::log('Pre check: maximum connections reached, quitting.', Logger::DEBUG);
return;
}
// Possibly there are too much database processes that block the system
if ($a->isMaxProcessesReached()) {
Logger::log('Pre check: maximum processes reached, quitting.', Logger::DEBUG);
return;
}
// Now we start additional cron processes if we should do so
if ($run_cron) {
self::runCron();
}
$starttime = time();
self::$mode = 1;
// We fetch the next queue entry that is about to be executed
2019-02-17 13:55:17 -05:00
while ($r = self::workerProcess()) {
foreach ($r as $entry) {
// Assure that the priority is an integer value
$entry['priority'] = (int)$entry['priority'];
// The work will be done
if (!self::execute($entry)) {
Logger::log('Process execution failed, quitting.', Logger::DEBUG);
return;
}
// If possible we will fetch new jobs for this worker
2019-02-17 13:55:17 -05:00
if (!self::getWaitingJobForPID() && Lock::acquire('worker_process', 0)) {
self::findWorkerProcesses();
Lock::release('worker_process');
self::$mode = 3;
}
}
self::$mode = 4;
// Quit the worker once every cron interval
if (time() > ($starttime + Config::get('system', 'cron_interval'))) {
Logger::log('Process lifetime reached, quitting.', Logger::DEBUG);
return;
}
2017-11-19 16:47:21 -05:00
// To avoid the quitting of multiple workers only one worker at a time will execute the check
if (Lock::acquire('worker', 0)) {
// Count active workers and compare them with a maximum value that depends on the load
2019-02-17 13:55:17 -05:00
if (self::tooMuchWorkers()) {
Logger::log('Active worker limit reached, quitting.', Logger::DEBUG);
Lock::release('worker');
return;
}
// Check free memory
if ($a->isMinMemoryReached()) {
Logger::log('Memory limit reached, quitting.', Logger::DEBUG);
Lock::release('worker');
return;
}
Lock::release('worker');
}
}
// Cleaning up. Possibly not needed, but it doesn't harm anything.
if (Config::get('system', 'worker_daemon_mode', false)) {
self::IPCSetJobState(false);
}
Logger::log("Couldn't select a workerqueue entry, quitting process " . getmypid() . ".", Logger::DEBUG);
}
2019-02-09 18:10:15 -05:00
/**
* @brief Check if non executed tasks do exist in the worker queue
*
* @return boolean Returns "true" if tasks are existing
* @throws \Exception
*/
private static function entriesExists()
{
$stamp = (float)microtime(true);
$exists = DBA::exists('workerqueue', ["NOT `done` AND `pid` = 0 AND `next_try` < ?", DateTimeFormat::utcNow()]);
self::$db_duration += (microtime(true) - $stamp);
return $exists;
}
/**
* @brief Returns the number of deferred entries in the worker queue
*
* @return integer Number of deferred entries in the worker queue
2019-01-06 16:06:53 -05:00
* @throws \Exception
*/
private static function deferredEntries()
{
2019-02-09 18:10:15 -05:00
$stamp = (float)microtime(true);
$count = DBA::count('workerqueue', ["NOT `done` AND `pid` = 0 AND `next_try` > ?", DateTimeFormat::utcNow()]);
self::$db_duration += (microtime(true) - $stamp);
self::$db_duration_count += (microtime(true) - $stamp);
return $count;
}
/**
* @brief Returns the number of non executed entries in the worker queue
*
* @return integer Number of non executed entries in the worker queue
2019-01-06 16:06:53 -05:00
* @throws \Exception
*/
private static function totalEntries()
{
2019-02-09 18:10:15 -05:00
$stamp = (float)microtime(true);
$count = DBA::count('workerqueue', ['done' => false, 'pid' => 0]);
self::$db_duration += (microtime(true) - $stamp);
self::$db_duration_count += (microtime(true) - $stamp);
return $count;
}
/**
* @brief Returns the highest priority in the worker queue that isn't executed
*
2017-11-19 16:47:21 -05:00
* @return integer Number of active worker processes
2019-01-06 16:06:53 -05:00
* @throws \Exception
*/
private static function highestPriority()
{
2019-02-09 18:10:15 -05:00
$stamp = (float)microtime(true);
$condition = ["`pid` = 0 AND NOT `done` AND `next_try` < ?", DateTimeFormat::utcNow()];
$workerqueue = DBA::selectFirst('workerqueue', ['priority'], $condition, ['order' => ['priority']]);
2019-02-09 18:10:15 -05:00
self::$db_duration += (microtime(true) - $stamp);
2018-07-21 08:46:04 -04:00
if (DBA::isResult($workerqueue)) {
return $workerqueue["priority"];
} else {
return 0;
}
}
/**
* @brief Returns if a process with the given priority is running
*
* @param integer $priority The priority that should be checked
*
* @return integer Is there a process running with that priority?
2019-01-06 16:06:53 -05:00
* @throws \Exception
*/
private static function processWithPriorityActive($priority)
{
$condition = ["`priority` <= ? AND `pid` != 0 AND NOT `done`", $priority];
return DBA::exists('workerqueue', $condition);
}
/**
* @brief Execute a worker entry
*
* @param array $queue Workerqueue entry
*
* @return boolean "true" if further processing should be stopped
2019-01-06 16:06:53 -05:00
* @throws \Friendica\Network\HTTPException\InternalServerErrorException
*/
public static function execute($queue)
{
$a = \get_app();
$mypid = getmypid();
// Quit when in maintenance
if (Config::get('system', 'maintenance', false, true)) {
Logger::log("Maintenance mode - quit process ".$mypid, Logger::DEBUG);
return false;
}
// Constantly check the number of parallel database processes
if ($a->isMaxProcessesReached()) {
Logger::log("Max processes reached for process ".$mypid, Logger::DEBUG);
return false;
}
// Constantly check the number of available database connections to let the frontend be accessible at any time
if (self::maxConnectionsReached()) {
Logger::log("Max connection reached for process ".$mypid, Logger::DEBUG);
return false;
}
$argv = json_decode($queue["parameter"], true);
// Check for existance and validity of the include file
$include = $argv[0];
if (method_exists(sprintf('Friendica\Worker\%s', $include), 'execute')) {
// We constantly update the "executed" date every minute to avoid being killed too soon
if (!isset(self::$last_update)) {
self::$last_update = strtotime($queue["executed"]);
}
$age = (time() - self::$last_update) / 60;
self::$last_update = time();
if ($age > 1) {
$stamp = (float)microtime(true);
DBA::update('workerqueue', ['executed' => DateTimeFormat::utcNow()], ['pid' => $mypid, 'done' => false]);
self::$db_duration += (microtime(true) - $stamp);
2019-02-09 18:10:15 -05:00
self::$db_duration_write += (microtime(true) - $stamp);
}
array_shift($argv);
self::execFunction($queue, $include, $argv, true);
$stamp = (float)microtime(true);
$condition = ["`id` = ? AND `next_try` < ?", $queue['id'], DateTimeFormat::utcNow()];
if (DBA::update('workerqueue', ['done' => true], $condition)) {
Config::set('system', 'last_worker_execution', DateTimeFormat::utcNow());
}
self::$db_duration = (microtime(true) - $stamp);
2019-02-09 18:10:15 -05:00
self::$db_duration_write += (microtime(true) - $stamp);
return true;
}
// The script could be provided as full path or only with the function name
if ($include == basename($include)) {
2017-11-14 17:13:33 -05:00
$include = "include/".$include.".php";
}
if (!validate_include($include)) {
2018-10-29 17:20:46 -04:00
Logger::log("Include file ".$argv[0]." is not valid!");
2019-02-09 18:10:15 -05:00
$stamp = (float)microtime(true);
DBA::delete('workerqueue', ['id' => $queue["id"]]);
2019-02-09 18:10:15 -05:00
self::$db_duration = (microtime(true) - $stamp);
self::$db_duration_write += (microtime(true) - $stamp);
return true;
}
require_once $include;
$funcname = str_replace(".php", "", basename($argv[0]))."_run";
if (function_exists($funcname)) {
// We constantly update the "executed" date every minute to avoid being killed too soon
if (!isset(self::$last_update)) {
self::$last_update = strtotime($queue["executed"]);
}
$age = (time() - self::$last_update) / 60;
self::$last_update = time();
if ($age > 1) {
$stamp = (float)microtime(true);
DBA::update('workerqueue', ['executed' => DateTimeFormat::utcNow()], ['pid' => $mypid, 'done' => false]);
self::$db_duration += (microtime(true) - $stamp);
2019-02-09 18:10:15 -05:00
self::$db_duration_write += (microtime(true) - $stamp);
}
self::execFunction($queue, $funcname, $argv, false);
$stamp = (float)microtime(true);
if (DBA::update('workerqueue', ['done' => true], ['id' => $queue["id"]])) {
Config::set('system', 'last_worker_execution', DateTimeFormat::utcNow());
}
self::$db_duration = (microtime(true) - $stamp);
2019-02-09 18:10:15 -05:00
self::$db_duration_write += (microtime(true) - $stamp);
} else {
2018-10-29 17:20:46 -04:00
Logger::log("Function ".$funcname." does not exist");
2019-02-09 18:10:15 -05:00
$stamp = (float)microtime(true);
DBA::delete('workerqueue', ['id' => $queue["id"]]);
2019-02-09 18:10:15 -05:00
self::$db_duration = (microtime(true) - $stamp);
self::$db_duration_write += (microtime(true) - $stamp);
}
return true;
}
/**
* @brief Execute a function from the queue
*
* @param array $queue Workerqueue entry
* @param string $funcname name of the function
* @param array $argv Array of values to be passed to the function
* @param boolean $method_call boolean
* @return void
2019-01-06 16:06:53 -05:00
* @throws \Friendica\Network\HTTPException\InternalServerErrorException
*/
private static function execFunction($queue, $funcname, $argv, $method_call)
{
$a = \get_app();
$argc = count($argv);
$logger = $a->getLogger();
$workerLogger = new WorkerLogger($logger, $funcname);
2019-02-22 14:41:13 -05:00
$workerLogger ->info("Process start.", ['priority' => $queue["priority"], 'id' => $queue["id"]]);
$stamp = (float)microtime(true);
// We use the callstack here to analyze the performance of executed worker entries.
// For this reason the variables have to be initialized.
2019-02-16 17:11:30 -05:00
$a->getProfiler()->reset();
$a->queue = $queue;
2019-02-09 18:10:15 -05:00
$up_duration = microtime(true) - self::$up_start;
// Reset global data to avoid interferences
unset($_SESSION);
2019-02-22 14:41:13 -05:00
// Set the workerLogger as new default logger
Logger::init($workerLogger);
if ($method_call) {
call_user_func_array(sprintf('Friendica\Worker\%s::execute', $funcname), $argv);
} else {
$funcname($argv, $argc);
}
Logger::init($logger);
unset($a->queue);
$duration = (microtime(true) - $stamp);
/* With these values we can analyze how effective the worker is.
* The database and rest time should be low since this is the unproductive time.
* The execution time is the productive time.
* By changing parameters like the maximum number of workers we can check the effectivness.
*/
$dbtotal = round(self::$db_duration, 2);
$dbread = round(self::$db_duration - (self::$db_duration_count + self::$db_duration_write + self::$db_duration_stat), 2);
$dbcount = round(self::$db_duration_count, 2);
$dbstat = round(self::$db_duration_stat, 2);
$dbwrite = round(self::$db_duration_write, 2);
$dblock = round(self::$lock_duration, 2);
$rest = round(max(0, $up_duration - (self::$db_duration + self::$lock_duration)), 2);
$exec = round($duration, 2);
2019-02-21 14:32:31 -05:00
$logger->info('Performance log.', ['mode' => self::$mode, 'count' => $dbcount, 'stat' => $dbstat, 'write' => $dbwrite, 'lock' => $dblock, 'total' => $dbtotal, 'rest' => $rest, 'exec' => $exec]);
2019-02-09 18:10:15 -05:00
self::$up_start = microtime(true);
self::$db_duration = 0;
self::$db_duration_count = 0;
self::$db_duration_stat = 0;
self::$db_duration_write = 0;
self::$lock_duration = 0;
self::$mode = 2;
if ($duration > 3600) {
2019-02-23 04:24:22 -05:00
$logger->info('Longer than 1 hour.', ['priority' => $queue["priority"], 'id' => $queue["id"], 'duration' => round($duration/60, 3)]);
} elseif ($duration > 600) {
2019-02-23 04:24:22 -05:00
$logger->info('Longer than 10 minutes.', ['priority' => $queue["priority"], 'id' => $queue["id"], 'duration' => round($duration/60, 3)]);
} elseif ($duration > 300) {
2019-02-23 04:24:22 -05:00
$logger->info('Longer than 5 minutes.', ['priority' => $queue["priority"], 'id' => $queue["id"], 'duration' => round($duration/60, 3)]);
} elseif ($duration > 120) {
2019-02-23 04:24:22 -05:00
$logger->info('Longer than 2 minutes.', ['priority' => $queue["priority"], 'id' => $queue["id"], 'duration' => round($duration/60, 3)]);
}
2019-02-22 14:41:13 -05:00
$workerLogger->info('Process done. ', ['priority' => $queue["priority"], 'id' => $queue["id"], 'duration' => number_format($duration, 4)]);
2019-02-18 05:18:05 -05:00
$a->getProfiler()->saveLog($a->getLogger(), "ID " . $queue["id"] . ": " . $funcname);
$cooldown = Config::get("system", "worker_cooldown", 0);
if ($cooldown > 0) {
2019-02-23 04:24:22 -05:00
$logger->info('Cooldown.', ['priority' => $queue["priority"], 'id' => $queue["id"], 'cooldown' => $cooldown]);
sleep($cooldown);
}
}
/**
* @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?
2019-01-06 16:06:53 -05:00
* @throws \Friendica\Network\HTTPException\InternalServerErrorException
*/
private static function maxConnectionsReached()
{
// Fetch the max value from the config. This is needed when the system cannot detect the correct value by itself.
$max = Config::get("system", "max_connections");
2017-11-19 16:47:21 -05:00
// Fetch the percentage level where the worker will get active
$maxlevel = Config::get("system", "max_connections_level", 75);
if ($max == 0) {
// the maximum number of possible user connections can be a system variable
$r = DBA::fetchFirst("SHOW VARIABLES WHERE `variable_name` = 'max_user_connections'");
2018-07-21 08:46:04 -04:00
if (DBA::isResult($r)) {
2017-11-05 12:13:37 -05:00
$max = $r["Value"];
}
// Or it can be granted. This overrides the system variable
2019-02-09 18:10:15 -05:00
$stamp = (float)microtime(true);
$r = DBA::p('SHOW GRANTS');
2019-02-09 18:10:15 -05:00
self::$db_duration += (microtime(true) - $stamp);
while ($grants = DBA::fetch($r)) {
2017-11-05 12:13:37 -05:00
$grant = array_pop($grants);
if (stristr($grant, "GRANT USAGE ON")) {
if (preg_match("/WITH MAX_USER_CONNECTIONS (\d*)/", $grant, $match)) {
$max = $match[1];
}
}
}
DBA::close($r);
}
// 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) {
2019-02-09 18:10:15 -05:00
$stamp = (float)microtime(true);
$r = DBA::p('SHOW PROCESSLIST');
2019-02-09 18:10:15 -05:00
self::$db_duration += (microtime(true) - $stamp);
2018-07-20 22:05:12 -04:00
$used = DBA::numRows($r);
DBA::close($r);
Logger::log("Connection usage (user values): ".$used."/".$max, Logger::DEBUG);
$level = ($used / $max) * 100;
if ($level >= $maxlevel) {
2018-10-29 17:20:46 -04:00
Logger::log("Maximum level (".$maxlevel."%) of user connections reached: ".$used."/".$max);
return true;
}
}
// We will now check for the system values.
// This limit could be reached although the user limits are fine.
$r = DBA::fetchFirst("SHOW VARIABLES WHERE `variable_name` = 'max_connections'");
2018-07-21 08:46:04 -04:00
if (!DBA::isResult($r)) {
return false;
}
2017-11-05 12:13:37 -05:00
$max = intval($r["Value"]);
if ($max == 0) {
return false;
}
$r = DBA::fetchFirst("SHOW STATUS WHERE `variable_name` = 'Threads_connected'");
2018-07-21 08:46:04 -04:00
if (!DBA::isResult($r)) {
return false;
}
2017-11-05 12:13:37 -05:00
$used = intval($r["Value"]);
if ($used == 0) {
return false;
}
Logger::log("Connection usage (system values): ".$used."/".$max, Logger::DEBUG);
$level = $used / $max * 100;
if ($level < $maxlevel) {
return false;
}
2018-10-29 17:20:46 -04:00
Logger::log("Maximum level (".$level."%) of system connections reached: ".$used."/".$max);
return true;
}
/**
* @brief fix the queue entry if the worker process died
* @return void
2019-01-06 16:06:53 -05:00
* @throws \Exception
*/
private static function killStaleWorkers()
{
2019-02-09 18:10:15 -05:00
$stamp = (float)microtime(true);
$entries = DBA::select(
'workerqueue',
['id', 'pid', 'executed', 'priority', 'parameter'],
['NOT `done` AND `pid` != 0'],
['order' => ['priority', 'created']]
);
2019-02-09 18:10:15 -05:00
self::$db_duration += (microtime(true) - $stamp);
while ($entry = DBA::fetch($entries)) {
if (!posix_kill($entry["pid"], 0)) {
2019-02-09 18:10:15 -05:00
$stamp = (float)microtime(true);
DBA::update(
'workerqueue',
['executed' => DBA::NULL_DATETIME, 'pid' => 0],
['id' => $entry["id"]]
);
2019-02-09 18:10:15 -05:00
self::$db_duration += (microtime(true) - $stamp);
self::$db_duration_write += (microtime(true) - $stamp);
} else {
// Kill long running processes
// Check if the priority is in a valid range
if (!in_array($entry["priority"], [PRIORITY_CRITICAL, PRIORITY_HIGH, PRIORITY_MEDIUM, PRIORITY_LOW, PRIORITY_NEGLIGIBLE])) {
$entry["priority"] = PRIORITY_MEDIUM;
}
// Define the maximum durations
$max_duration_defaults = [PRIORITY_CRITICAL => 720, PRIORITY_HIGH => 10, PRIORITY_MEDIUM => 60, PRIORITY_LOW => 180, PRIORITY_NEGLIGIBLE => 720];
$max_duration = $max_duration_defaults[$entry["priority"]];
$argv = json_decode($entry["parameter"], true);
$argv[0] = basename($argv[0]);
// How long is the process already running?
$duration = (time() - strtotime($entry["executed"])) / 60;
if ($duration > $max_duration) {
2018-10-29 17:20:46 -04:00
Logger::log("Worker process ".$entry["pid"]." (".substr(json_encode($argv), 0, 50).") took more than ".$max_duration." minutes. It will be killed now.");
posix_kill($entry["pid"], SIGTERM);
// We killed the stale process.
// To avoid a blocking situation we reschedule the process at the beginning of the queue.
// Additionally we are lowering the priority. (But not PRIORITY_CRITICAL)
$new_priority = $entry["priority"];
if ($entry["priority"] == PRIORITY_HIGH) {
$new_priority = PRIORITY_MEDIUM;
} elseif ($entry["priority"] == PRIORITY_MEDIUM) {
$new_priority = PRIORITY_LOW;
} elseif ($entry["priority"] != PRIORITY_CRITICAL) {
$new_priority = PRIORITY_NEGLIGIBLE;
}
2019-02-09 18:10:15 -05:00
$stamp = (float)microtime(true);
DBA::update(
'workerqueue',
['executed' => DBA::NULL_DATETIME, 'created' => DateTimeFormat::utcNow(), 'priority' => $new_priority, 'pid' => 0],
['id' => $entry["id"]]
);
2019-02-09 18:10:15 -05:00
self::$db_duration += (microtime(true) - $stamp);
self::$db_duration_write += (microtime(true) - $stamp);
} else {
Logger::log("Worker process ".$entry["pid"]." (".substr(json_encode($argv), 0, 50).") now runs for ".round($duration)." of ".$max_duration." allowed minutes. That's okay.", Logger::DEBUG);
}
}
}
}
/**
* @brief Checks if the number of active workers exceeds the given limits
*
* @return bool Are there too much workers running?
2019-01-06 16:06:53 -05:00
* @throws \Friendica\Network\HTTPException\InternalServerErrorException
*/
2019-02-17 13:55:17 -05:00
private static function tooMuchWorkers()
{
$queues = Config::get("system", "worker_queues", 4);
$maxqueues = $queues;
$active = self::activeWorkers();
// Decrease the number of workers at higher load
$load = System::currentLoad();
if ($load) {
$maxsysload = intval(Config::get("system", "maxloadavg", 50));
/* Default exponent 3 causes queues to rapidly decrease as load increases.
* If you have 20 max queues at idle, then you get only 5 queues at 37.1% of $maxsysload.
* For some environments, this rapid decrease is not needed.
* With exponent 1, you could have 20 max queues at idle and 13 at 37% of $maxsysload.
*/
2018-06-20 06:22:53 -04:00
$exponent = intval(Config::get('system', 'worker_load_exponent', 3));
2018-06-20 06:06:20 -04:00
$slope = pow(max(0, $maxsysload - $load) / $maxsysload, $exponent);
$queues = intval(ceil($slope * $maxqueues));
$processlist = '';
if (Config::get('system', 'worker_jpm')) {
2019-02-09 18:10:15 -05:00
$intervals = explode(',', Config::get('system', 'worker_jpm_range'));
$jobs_per_minute = [];
foreach ($intervals as $interval) {
2019-02-10 23:39:24 -05:00
if ($interval == 0) {
continue;
} else {
$interval = (int)$interval;
}
2019-02-09 18:10:15 -05:00
$stamp = (float)microtime(true);
$jobs = DBA::p("SELECT COUNT(*) AS `jobs` FROM `workerqueue` WHERE `done` AND `executed` > UTC_TIMESTAMP() - INTERVAL ? MINUTE", $interval);
2019-02-09 18:10:15 -05:00
self::$db_duration += (microtime(true) - $stamp);
2019-02-17 14:20:24 -05:00
self::$db_duration_stat += (microtime(true) - $stamp);
if ($job = DBA::fetch($jobs)) {
$jobs_per_minute[$interval] = number_format($job['jobs'] / $interval, 0);
}
DBA::close($jobs);
}
$processlist = ' - jpm: '.implode('/', $jobs_per_minute);
}
2019-02-09 18:10:15 -05:00
// Create a list of queue entries grouped by their priority
$listitem = [0 => ''];
2019-02-09 18:10:15 -05:00
$idle_workers = $active;
2019-02-17 13:55:17 -05:00
$deferred = self::deferredEntries();
2019-02-11 03:59:14 -05:00
2019-02-09 18:10:15 -05:00
if (Config::get('system', 'worker_debug')) {
2019-02-11 03:59:14 -05:00
$waiting_processes = 0;
// Now adding all processes with workerqueue entries
2019-02-09 18:10:15 -05:00
$stamp = (float)microtime(true);
$jobs = DBA::p("SELECT COUNT(*) AS `entries`, `priority` FROM `workerqueue` WHERE NOT `done` AND `next_try` < ? GROUP BY `priority`", DateTimeFormat::utcNow());
self::$db_duration += (microtime(true) - $stamp);
self::$db_duration_stat += (microtime(true) - $stamp);
while ($entry = DBA::fetch($jobs)) {
$stamp = (float)microtime(true);
$processes = DBA::p("SELECT COUNT(*) AS `running` FROM `process` INNER JOIN `workerqueue` ON `workerqueue`.`pid` = `process`.`pid` WHERE NOT `done` AND `priority` = ?", $entry["priority"]);
2019-02-09 18:10:15 -05:00
self::$db_duration += (microtime(true) - $stamp);
self::$db_duration_stat += (microtime(true) - $stamp);
if ($process = DBA::fetch($processes)) {
$idle_workers -= $process["running"];
2019-02-11 03:59:14 -05:00
$waiting_processes += $entry["entries"];
$listitem[$entry["priority"]] = $entry["priority"].":".$process["running"]."/".$entry["entries"];
}
DBA::close($processes);
}
2019-02-09 18:10:15 -05:00
DBA::close($jobs);
2019-02-17 13:55:17 -05:00
$entries = $deferred + $waiting_processes;
2019-02-09 18:10:15 -05:00
} else {
2019-02-17 13:55:17 -05:00
$entries = self::totalEntries();
$waiting_processes = max(0, $entries - $deferred);
2019-02-09 18:10:15 -05:00
$stamp = (float)microtime(true);
$jobs = DBA::p("SELECT COUNT(*) AS `running`, `priority` FROM `process` INNER JOIN `workerqueue` ON `workerqueue`.`pid` = `process`.`pid` AND NOT `done` GROUP BY `priority` ORDER BY `priority`");
self::$db_duration += (microtime(true) - $stamp);
2019-02-17 13:55:17 -05:00
self::$db_duration_stat += (microtime(true) - $stamp);
2019-02-09 18:10:15 -05:00
while ($entry = DBA::fetch($jobs)) {
$idle_workers -= $entry["running"];
$listitem[$entry["priority"]] = $entry["priority"].":".$entry["running"];
}
DBA::close($jobs);
}
2019-02-09 18:10:15 -05:00
$listitem[0] = "0:" . max(0, $idle_workers);
$processlist .= ' ('.implode(', ', $listitem).')';
2019-02-09 18:10:15 -05:00
if (Config::get("system", "worker_fastlane", false) && ($queues > 0) && self::entriesExists() && ($active >= $queues)) {
$top_priority = self::highestPriority();
$high_running = self::processWithPriorityActive($top_priority);
if (!$high_running && ($top_priority > PRIORITY_UNDEFINED) && ($top_priority < PRIORITY_NEGLIGIBLE)) {
Logger::log("There are jobs with priority ".$top_priority." waiting but none is executed. Open a fastlane.", Logger::DEBUG);
$queues = $active + 1;
}
}
2019-02-11 03:59:14 -05:00
Logger::log("Load: " . $load ."/" . $maxsysload . " - processes: " . $deferred . "/" . $active . "/" . $waiting_processes . $processlist . " - maximum: " . $queues . "/" . $maxqueues, Logger::DEBUG);
// Are there fewer workers running as possible? Then fork a new one.
if (!Config::get("system", "worker_dont_fork", false) && ($queues > ($active + 1)) && ($entries > 1)) {
Logger::log("Active workers: ".$active."/".$queues." Fork a new worker.", Logger::DEBUG);
if (Config::get('system', 'worker_daemon_mode', false)) {
self::IPCSetJobState(true);
} else {
self::spawnWorker();
}
}
}
// if there are too much worker, we don't spawn a new one.
if (Config::get('system', 'worker_daemon_mode', false) && ($active > $queues)) {
self::IPCSetJobState(false);
}
return $active > $queues;
}
/**
2017-11-19 16:47:21 -05:00
* @brief Returns the number of active worker processes
*
2017-11-19 16:47:21 -05:00
* @return integer Number of active worker processes
2019-01-06 16:06:53 -05:00
* @throws \Exception
*/
private static function activeWorkers()
{
2019-02-09 18:10:15 -05:00
$stamp = (float)microtime(true);
$count = DBA::count('process', ['command' => 'Worker.php']);
self::$db_duration += (microtime(true) - $stamp);
return $count;
}
2019-02-17 14:20:24 -05:00
/**
* @brief Returns waiting jobs for the current process id
*
* @return array waiting workerqueue jobs
* @throws \Exception
*/
2019-02-17 13:55:17 -05:00
private static function getWaitingJobForPID()
{
$stamp = (float)microtime(true);
$r = DBA::select('workerqueue', [], ['pid' => getmypid(), 'done' => false]);
self::$db_duration += (microtime(true) - $stamp);
if (DBA::isResult($r)) {
return DBA::toArray($r);
}
DBA::close($r);
return false;
}
2019-02-17 14:20:24 -05:00
/**
* @brief Returns the next jobs that should be executed
*
* @return array array with next jobs
* @throws \Exception
*/
2019-02-17 13:55:17 -05:00
private static function nextProcess()
2019-02-16 10:03:37 -05:00
{
$priority = self::nextPriority();
if (empty($priority)) {
2019-02-21 14:32:31 -05:00
Logger::info('No tasks found');
2019-02-16 10:03:37 -05:00
return [];
}
if ($priority <= PRIORITY_MEDIUM) {
$limit = Config::get('system', 'worker_fetch_limit', 1);
} else {
$limit = 1;
}
$ids = [];
$stamp = (float)microtime(true);
$condition = ["`priority` = ? AND `pid` = 0 AND NOT `done` AND `next_try` < ?", $priority, DateTimeFormat::utcNow()];
$tasks = DBA::select('workerqueue', ['id'], $condition, ['limit' => $limit, 'order' => ['created']]);
self::$db_duration += (microtime(true) - $stamp);
while ($task = DBA::fetch($tasks)) {
$ids[] = $task['id'];
}
DBA::close($tasks);
2019-02-21 14:32:31 -05:00
Logger::info('Found:', ['id' => $ids, 'priority' => $priority]);
2019-02-16 10:03:37 -05:00
return $ids;
}
2019-02-17 14:20:24 -05:00
/**
* @brief Returns the priority of the next workerqueue job
*
* @return string priority
* @throws \Exception
*/
2019-02-17 13:55:17 -05:00
private static function nextPriority()
2019-02-16 10:03:37 -05:00
{
$waiting = [];
$priorities = [PRIORITY_CRITICAL, PRIORITY_HIGH, PRIORITY_MEDIUM, PRIORITY_LOW, PRIORITY_NEGLIGIBLE];
foreach ($priorities as $priority) {
$stamp = (float)microtime(true);
if (DBA::exists('workerqueue', ["`priority` = ? AND `pid` = 0 AND NOT `done` AND `next_try` < ?", $priority, DateTimeFormat::utcNow()])) {
$waiting[$priority] = true;
}
self::$db_duration += (microtime(true) - $stamp);
}
if (!empty($waiting[PRIORITY_CRITICAL])) {
return PRIORITY_CRITICAL;
}
$running = [];
$running_total = 0;
2019-02-16 10:03:37 -05:00
$stamp = (float)microtime(true);
$processes = DBA::p("SELECT COUNT(DISTINCT(`process`.`pid`)) AS `running`, `priority` FROM `process`
INNER JOIN `workerqueue` ON `workerqueue`.`pid` = `process`.`pid`
WHERE NOT `done` GROUP BY `priority`");
self::$db_duration += (microtime(true) - $stamp);
while ($process = DBA::fetch($processes)) {
$running[$process['priority']] = $process['running'];
$running_total += $process['running'];
2019-02-16 10:03:37 -05:00
}
DBA::close($processes);
foreach ($priorities as $priority) {
if (!empty($waiting[$priority]) && empty($running[$priority])) {
2019-02-21 14:32:31 -05:00
Logger::info('No running worker found with priority {priority} - assigning it.', ['priority' => $priority]);
2019-02-16 10:03:37 -05:00
return $priority;
}
}
$active = max(self::activeWorkers(), $running_total);
$priorities = max(count($waiting), count($running));
$exponent = 2;
$total = 0;
for ($i = 1; $i <= $priorities; ++$i) {
$total += pow($i, $exponent);
2019-02-16 10:03:37 -05:00
}
$limit = [];
for ($i = 1; $i <= $priorities; ++$i) {
$limit[$priorities - $i] = max(1, round($active * (pow($i, $exponent) / $total)));
}
$i = 0;
foreach ($running as $priority => $workers) {
if ($workers < $limit[$i++]) {
2019-02-21 14:32:31 -05:00
Logger::info('Priority {priority} has got {workers} workers out of a limit of {limit}', ['priority' => $priority, 'workers' => $workers, 'limit' => $limit[$i - 1]]);
return $priority;
}
2019-02-16 10:03:37 -05:00
}
if (!empty($waiting)) {
$priority = array_keys($waiting)[0];
2019-02-21 14:32:31 -05:00
Logger::info('No underassigned priority found, now taking the highest priority.', ['priority' => $priority]);
return $priority;
2019-02-16 10:03:37 -05:00
}
return false;
}
/**
* @brief Find and claim the next worker process for us
*
* @return boolean Have we found something?
2019-01-06 16:06:53 -05:00
* @throws \Friendica\Network\HTTPException\InternalServerErrorException
*/
2019-02-17 13:55:17 -05:00
private static function findWorkerProcesses()
{
$mypid = getmypid();
2019-02-16 10:03:37 -05:00
$ids = self::nextProcess();
2019-02-17 13:55:17 -05:00
// If there is no result we check without priority limit
if (empty($ids)) {
2019-02-09 18:10:15 -05:00
$stamp = (float)microtime(true);
2019-02-17 14:20:24 -05:00
$condition = ["`pid` = 0 AND NOT `done` AND `next_try` < ?", DateTimeFormat::utcNow()];
$result = DBA::select('workerqueue', ['id'], $condition, ['limit' => 1, 'order' => ['priority', 'created']]);
2019-02-09 18:10:15 -05:00
self::$db_duration += (microtime(true) - $stamp);
while ($id = DBA::fetch($result)) {
$ids[] = $id["id"];
}
DBA::close($result);
}
2019-02-17 13:55:17 -05:00
if (!empty($ids)) {
2019-02-09 18:10:15 -05:00
$stamp = (float)microtime(true);
2019-02-17 14:20:24 -05:00
$condition = ['id' => $ids, 'done' => false, 'pid' => 0];
DBA::update('workerqueue', ['executed' => DateTimeFormat::utcNow(), 'pid' => $mypid], $condition);
2019-02-09 18:10:15 -05:00
self::$db_duration += (microtime(true) - $stamp);
self::$db_duration_write += (microtime(true) - $stamp);
}
2019-02-17 13:55:17 -05:00
return !empty($ids);
}
/**
* @brief Returns the next worker process
*
* @return string SQL statement
2019-01-06 16:06:53 -05:00
* @throws \Friendica\Network\HTTPException\InternalServerErrorException
*/
2019-02-17 13:55:17 -05:00
public static function workerProcess()
{
// There can already be jobs for us in the queue.
2019-02-17 13:55:17 -05:00
$waiting = self::getWaitingJobForPID();
if (!empty($waiting)) {
return $waiting;
}
2019-02-09 18:10:15 -05:00
$stamp = (float)microtime(true);
if (!Lock::acquire('worker_process')) {
return false;
}
2019-02-09 18:10:15 -05:00
self::$lock_duration += (microtime(true) - $stamp);
2019-02-17 13:55:17 -05:00
$found = self::findWorkerProcesses();
Lock::release('worker_process');
if ($found) {
2019-02-09 18:10:15 -05:00
$stamp = (float)microtime(true);
$r = DBA::select('workerqueue', [], ['pid' => getmypid(), 'done' => false]);
2019-02-09 18:10:15 -05:00
self::$db_duration += (microtime(true) - $stamp);
2018-07-20 22:03:40 -04:00
return DBA::toArray($r);
}
2017-11-05 12:13:37 -05:00
return false;
}
/**
* @brief Removes a workerqueue entry from the current process
* @return void
2019-01-06 16:06:53 -05:00
* @throws \Exception
*/
public static function unclaimProcess()
{
$mypid = getmypid();
2019-02-09 18:10:15 -05:00
$stamp = (float)microtime(true);
DBA::update('workerqueue', ['executed' => DBA::NULL_DATETIME, 'pid' => 0], ['pid' => $mypid, 'done' => false]);
2019-02-09 18:10:15 -05:00
self::$db_duration += (microtime(true) - $stamp);
self::$db_duration_write += (microtime(true) - $stamp);
}
/**
* @brief Call the front end worker
* @return void
2019-01-06 16:06:53 -05:00
* @throws \Friendica\Network\HTTPException\InternalServerErrorException
*/
public static function callWorker()
{
if (!Config::get("system", "frontend_worker")) {
return;
}
$url = System::baseUrl()."/worker";
Network::fetchUrl($url, false, $redirects, 1);
}
/**
* @brief Call the front end worker if there aren't any active
* @return void
2019-01-06 16:06:53 -05:00
* @throws \Friendica\Network\HTTPException\InternalServerErrorException
*/
public static function executeIfIdle()
{
if (!Config::get("system", "frontend_worker")) {
return;
}
2017-11-19 16:47:21 -05:00
// Do we have "proc_open"? Then we can fork the worker
if (function_exists("proc_open")) {
// When was the last time that we called the worker?
// Less than one minute? Then we quit
if ((time() - Config::get("system", "worker_started")) < 60) {
return;
}
Config::set("system", "worker_started", time());
// Do we have enough running workers? Then we quit here.
if (self::tooMuchWorkers()) {
// Cleaning dead processes
self::killStaleWorkers();
Process::deleteInactive();
return;
}
self::runCron();
Logger::log('Call worker', Logger::DEBUG);
2017-11-05 10:28:55 -05:00
self::spawnWorker();
return;
}
// We cannot execute background processes.
// We now run the processes from the frontend.
// This won't work with long running processes.
self::runCron();
self::clearProcesses();
2018-06-19 17:33:07 -04:00
$workers = self::activeWorkers();
2018-06-19 17:33:07 -04:00
if ($workers == 0) {
self::callWorker();
}
}
/**
* @brief Removes long running worker processes
* @return void
2019-01-06 16:06:53 -05:00
* @throws \Friendica\Network\HTTPException\InternalServerErrorException
*/
public static function clearProcesses()
{
$timeout = Config::get("system", "frontend_worker_timeout", 10);
/// @todo We should clean up the corresponding workerqueue entries as well
2019-02-09 18:10:15 -05:00
$stamp = (float)microtime(true);
$condition = ["`created` < ? AND `command` = 'worker.php'",
DateTimeFormat::utc("now - ".$timeout." minutes")];
DBA::delete('process', $condition);
2019-02-09 18:10:15 -05:00
self::$db_duration = (microtime(true) - $stamp);
self::$db_duration_write += (microtime(true) - $stamp);
}
/**
* @brief Runs the cron processes
* @return void
2019-01-06 16:06:53 -05:00
* @throws \Friendica\Network\HTTPException\InternalServerErrorException
*/
private static function runCron()
{
Logger::log('Add cron entries', Logger::DEBUG);
// Check for spooled items
self::add(['priority' => PRIORITY_HIGH, 'force_priority' => true], 'SpoolPost');
// Run the cron job that calls all other jobs
self::add(['priority' => PRIORITY_MEDIUM, 'force_priority' => true], 'Cron');
// Cleaning dead processes
self::killStaleWorkers();
}
/**
2018-05-14 01:02:18 -04:00
* @brief Spawns a new worker
2019-01-06 16:06:53 -05:00
* @param bool $do_cron
* @return void
2019-01-06 16:06:53 -05:00
* @throws \Friendica\Network\HTTPException\InternalServerErrorException
*/
public static function spawnWorker($do_cron = false)
{
$command = 'bin/worker.php';
$args = ['no_cron' => !$do_cron];
get_app()->proc_run($command, $args);
// after spawning we have to remove the flag.
if (Config::get('system', 'worker_daemon_mode', false)) {
self::IPCSetJobState(false);
}
2017-11-05 10:28:55 -05:00
}
/**
* @brief Adds tasks to the worker queue
*
2017-11-06 10:38:15 -05:00
* @param (integer|array) priority or parameter array, strings are deprecated and are ignored
*
* next args are passed as $cmd command line
* or: Worker::add(PRIORITY_HIGH, "Notifier", "drop", $drop_id);
2017-11-19 14:47:04 -05:00
* or: Worker::add(array('priority' => PRIORITY_HIGH, 'dont_fork' => true), "CreateShadowEntry", $post_id);
*
2019-01-06 16:06:53 -05:00
* @return boolean "false" if proc_run couldn't be executed
* @throws \Friendica\Network\HTTPException\InternalServerErrorException
* @note $cmd and string args are surrounded with ""
*
* @hooks 'proc_run'
2019-01-06 16:06:53 -05:00
* array $arr
*
*/
public static function add($cmd)
{
$args = func_get_args();
if (!count($args)) {
return false;
}
$arr = ['args' => $args, 'run_cmd' => true];
Hook::callAll("proc_run", $arr);
if (!$arr['run_cmd'] || !count($args)) {
return true;
}
$priority = PRIORITY_MEDIUM;
$dont_fork = Config::get("system", "worker_dont_fork", false);
$created = DateTimeFormat::utcNow();
$force_priority = false;
$run_parameter = array_shift($args);
if (is_int($run_parameter)) {
$priority = $run_parameter;
} elseif (is_array($run_parameter)) {
if (isset($run_parameter['priority'])) {
$priority = $run_parameter['priority'];
}
if (isset($run_parameter['created'])) {
$created = $run_parameter['created'];
}
if (isset($run_parameter['dont_fork'])) {
$dont_fork = $run_parameter['dont_fork'];
}
if (isset($run_parameter['force_priority'])) {
$force_priority = $run_parameter['force_priority'];
}
}
$parameters = json_encode($args);
$found = DBA::exists('workerqueue', ['parameter' => $parameters, 'done' => false]);
// Quit if there was a database error - a precaution for the update process to 3.5.3
if (DBA::errorNo() != 0) {
return false;
}
if (!$found) {
DBA::insert('workerqueue', ['parameter' => $parameters, 'created' => $created, 'priority' => $priority]);
} elseif ($force_priority) {
DBA::update('workerqueue', ['priority' => $priority], ['parameter' => $parameters, 'done' => false, 'pid' => 0]);
}
2018-06-02 01:17:32 -04:00
// Should we quit and wait for the worker to be called as a cronjob?
if ($dont_fork) {
return true;
}
// If there is a lock then we don't have to check for too much worker
if (!Lock::acquire('worker', 0)) {
return true;
}
// If there are already enough workers running, don't fork another one
$quit = self::tooMuchWorkers();
Lock::release('worker');
if ($quit) {
return true;
}
// We tell the daemon that a new job entry exists
if (Config::get('system', 'worker_daemon_mode', false)) {
// We don't have to set the IPC flag - this is done in "tooMuchWorkers"
return true;
}
2017-11-19 16:47:21 -05:00
// Now call the worker to execute the jobs that we just added to the queue
2017-11-05 10:28:55 -05:00
self::spawnWorker();
return true;
}
2018-10-15 01:19:35 -04:00
/**
* Defers the current worker entry
*/
public static function defer()
{
if (empty(BaseObject::getApp()->queue)) {
return;
}
$queue = BaseObject::getApp()->queue;
$retrial = $queue['retrial'];
$id = $queue['id'];
$priority = $queue['priority'];
2018-10-15 01:19:35 -04:00
if ($retrial > 14) {
Logger::log('Id ' . $id . ' had been tried 14 times. We stop now.', Logger::DEBUG);
return;
2018-10-15 01:19:35 -04:00
}
// Calculate the delay until the next trial
$delay = (($retrial + 3) ** 4) + (rand(1, 30) * ($retrial + 1));
$next = DateTimeFormat::utc('now + ' . $delay . ' seconds');
if (($priority < PRIORITY_MEDIUM) && ($retrial > 2)) {
$priority = PRIORITY_MEDIUM;
} elseif (($priority < PRIORITY_LOW) && ($retrial > 5)) {
$priority = PRIORITY_LOW;
} elseif (($priority < PRIORITY_NEGLIGIBLE) && ($retrial > 7)) {
$priority = PRIORITY_NEGLIGIBLE;
}
Logger::log('Defer execution ' . $retrial . ' of id ' . $id . ' to ' . $next . ' - priority old/new: ' . $queue['priority'] . '/' . $priority, Logger::DEBUG);
2018-10-15 01:19:35 -04:00
2019-02-09 18:10:15 -05:00
$stamp = (float)microtime(true);
$fields = ['retrial' => $retrial + 1, 'next_try' => $next, 'executed' => DBA::NULL_DATETIME, 'pid' => 0, 'priority' => $priority];
2018-10-15 01:19:35 -04:00
DBA::update('workerqueue', $fields, ['id' => $id]);
2019-02-09 18:10:15 -05:00
self::$db_duration += (microtime(true) - $stamp);
self::$db_duration_write += (microtime(true) - $stamp);
2018-10-15 01:19:35 -04:00
}
/**
* Log active processes into the "process" table
*
* @brief Log active processes into the "process" table
*/
public static function startProcess()
{
$trace = debug_backtrace(DEBUG_BACKTRACE_IGNORE_ARGS, 1);
$command = basename($trace[0]['file']);
Process::deleteInactive();
Process::insert($command);
}
/**
* Remove the active process from the "process" table
*
* @brief Remove the active process from the "process" table
* @return bool
2019-01-06 16:06:53 -05:00
* @throws \Exception
*/
public static function endProcess()
{
return Process::deleteByPid();
}
/**
* Set the flag if some job is waiting
*
* @brief Set the flag if some job is waiting
* @param boolean $jobs Is there a waiting job?
2019-01-06 16:06:53 -05:00
* @throws \Exception
*/
public static function IPCSetJobState($jobs)
{
2019-02-09 18:10:15 -05:00
$stamp = (float)microtime(true);
DBA::update('worker-ipc', ['jobs' => $jobs], ['key' => 1], true);
2019-02-09 18:10:15 -05:00
self::$db_duration += (microtime(true) - $stamp);
self::$db_duration_write += (microtime(true) - $stamp);
}
/**
* Checks if some worker job waits to be executed
*
* @brief Checks if some worker job waits to be executed
* @return bool
2019-01-06 16:06:53 -05:00
* @throws \Exception
*/
public static function IPCJobsExists()
{
2019-02-09 18:10:15 -05:00
$stamp = (float)microtime(true);
$row = DBA::selectFirst('worker-ipc', ['jobs'], ['key' => 1]);
2019-02-09 18:10:15 -05:00
self::$db_duration += (microtime(true) - $stamp);
// When we don't have a row, no job is running
2018-07-21 08:46:04 -04:00
if (!DBA::isResult($row)) {
return false;
}
return (bool)$row['jobs'];
}
}