require_once("boot.php");
-
function poller_run(&$argv, &$argc){
global $a, $db;
}
if(is_null($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');
- require_once('mod/nodeinfo.php');
-
- load_config('config');
- load_config('system');
-
- $maxsysload = intval(get_config('system','maxloadavg'));
- if($maxsysload < 1)
- $maxsysload = 50;
+ @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);
+ };
+
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.');
+ logger('system: load ' . $load[0] . ' too high. poller deferred to next scheduled run.');
return;
}
}
- $lockpath = get_lockpath();
- if ($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;
- }
- }
-
-
-
- $a->set_baseurl(get_config('system','url'));
-
- load_hooks();
-
- logger('poller: start');
-
- // run queue delivery process in the background
-
- proc_run('php',"include/queue.php");
-
- // run diaspora photo queue process in the background
-
- proc_run('php',"include/dsprphotoq.php");
-
- // run the process to discover global contacts in the background
-
- proc_run('php',"include/discover_poco.php");
+ if(($argc <= 1) OR ($argv[1] != "no_cron")) {
+ // Run the cron job that calls all other jobs
+ proc_run("php","include/cron.php");
+
+ // Run the cronhooks job separately from cron for being able to use a different timing
+ proc_run("php","include/cronhooks.php");
+
+ // Cleaning dead processes
+ $r = q("SELECT DISTINCT(`pid`) FROM `workerqueue` WHERE `executed` != '0000-00-00 00:00:00'");
+ foreach($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 {
+ // To-Do: Kill long running processes
+ // But: Update processes (like the database update) mustn't be killed
+ }
- // run the process to update locally stored global contacts in the background
+ } else
+ // Sleep two seconds before checking for running processes to avoid having too many workers
+ sleep(4);
- proc_run('php',"include/discover_poco.php", "checkcontact");
+ // Checking number of workers
+ if (poller_too_much_workers())
+ return;
- // expire any expired accounts
+ $starttime = time();
- q("UPDATE user SET `account_expired` = 1 where `account_expired` = 0
- AND `account_expires_on` != '0000-00-00 00:00:00'
- AND `account_expires_on` < UTC_TIMESTAMP() ");
+ while ($r = q("SELECT * FROM `workerqueue` WHERE `executed` = '0000-00-00 00:00:00' ORDER BY `created` LIMIT 1")) {
- // delete user and contact records for recently removed accounts
+ 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"]));
- $r = q("SELECT * FROM `user` WHERE `account_removed` = 1 AND `account_expires_on` < UTC_TIMESTAMP() - INTERVAL 3 DAY");
- if ($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']));
+ // 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;
-
- // Check OStatus conversations
- // Check only conversations with mentions (for a longer time)
- check_conversations(true);
-
- // Check every conversation
- check_conversations(false);
-
- // Follow your friends from your legacy OStatus account
- ostatus_check_follow_friends();
-
- // update nodeinfo data
- nodeinfo_cron();
-
- // To-Do: Regenerate usage statistics
- // q("ANALYZE TABLE `item`");
-
- // once daily run birthday_updates and then expire in background
-
- $d1 = get_config('system','last_expire_day');
- $d2 = intval(datetime_convert('UTC','UTC','now','d'));
-
- if($d2 != intval($d1)) {
-
- update_contact_birthdays();
-
- update_suggestions();
- set_config('system','last_expire_day',$d2);
- proc_run('php','include/expire.php');
- }
-
- $last = get_config('system','cache_last_cleared');
-
- if($last) {
- $next = $last + (3600); // Once per hour
- $clear_cache = ($next <= time());
- } else
- $clear_cache = true;
-
- if ($clear_cache) {
- // clear old cache
- Cache::clear();
+ $argv = json_decode($r[0]["parameter"]);
- // clear old item cache files
- clear_cache();
+ $argc = count($argv);
- // clear cache for photos
- clear_cache($a->get_basepath(), $a->get_basepath()."/photo");
+ // Check for existance and validity of the include file
+ $include = $argv[0];
- // clear smarty cache
- clear_cache($a->get_basepath()."/view/smarty3/compiled", $a->get_basepath()."/view/smarty3/compiled");
-
- // clear cache for image proxy
- if (!get_config("system", "proxy_disabled")) {
- clear_cache($a->get_basepath(), $a->get_basepath()."/proxy");
-
- $cachetime = get_config('system','proxy_cache_time');
- if (!$cachetime) $cachetime = PROXY_DEFAULT_TIME;
-
- q('DELETE FROM `photo` WHERE `uid` = 0 AND `resource-id` LIKE "pic:%%" AND `created` < NOW() - INTERVAL %d SECOND', $cachetime);
+ if (!validate_include($include)) {
+ logger("Include file ".$argv[0]." is not valid!");
+ q("DELETE FROM `workerqueue` WHERE `id` = %d", intval($r[0]["id"]));
+ continue;
}
- set_config('system','cache_last_cleared', time());
- }
-
- $manual_id = 0;
- $generation = 0;
- $force = false;
- $restart = false;
-
- if(($argc > 1) && ($argv[1] == 'force'))
- $force = true;
-
- if(($argc > 1) && ($argv[1] == 'restart')) {
- $restart = true;
- $generation = intval($argv[2]);
- if(! $generation)
- killme();
- }
-
- if(($argc > 1) && intval($argv[1])) {
- $manual_id = intval($argv[1]);
- $force = true;
- }
-
- $interval = intval(get_config('system','poll_interval'));
- if(! $interval)
- $interval = ((get_config('system','delivery_interval') === false) ? 3 : intval(get_config('system','delivery_interval')));
+ require_once($include);
- $sql_extra = (($manual_id) ? " AND `id` = $manual_id " : "");
+ $funcname=str_replace(".php", "", basename($argv[0]))."_run";
- reload_plugins();
+ if (function_exists($funcname)) {
+ logger("Process ".getmypid().": ".$funcname." ".$r[0]["parameter"]);
+ $funcname($argv, $argc);
- $d = datetime_convert();
+ logger("Process ".getmypid().": ".$funcname." - done");
- if(! $restart)
- proc_run('php','include/cronhooks.php');
+ q("DELETE FROM `workerqueue` WHERE `id` = %d", intval($r[0]["id"]));
+ } else
+ logger("Function ".$funcname." does not exist");
- // 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.
-
- $abandon_sql = (($abandon_days)
- ? sprintf(" AND `user`.`login_date` > UTC_TIMESTAMP() - INTERVAL %d DAY ", intval($abandon_days))
- : ''
- );
-
- $contacts = q("SELECT `contact`.`id` FROM `contact` INNER JOIN `user` ON `user`.`uid` = `contact`.`uid`
- WHERE `rel` IN (%d, %d) AND `poll` != '' AND `network` IN ('%s', '%s', '%s', '%s', '%s', '%s')
- $sql_extra
- AND NOT `self` AND NOT `contact`.`blocked` AND NOT `contact`.`readonly` AND NOT `contact`.`archive`
- AND NOT `user`.`account_expired` AND NOT `user`.`account_removed` $abandon_sql ORDER BY RAND()",
- intval(CONTACT_IS_SHARING),
- intval(CONTACT_IS_FRIEND),
- dbesc(NETWORK_DFRN),
- dbesc(NETWORK_ZOT),
- dbesc(NETWORK_OSTATUS),
- dbesc(NETWORK_FEED),
- dbesc(NETWORK_MAIL),
- dbesc(NETWORK_MAIL2)
- );
+ // Quit the poller once every hour
+ if (time() > ($starttime + 3600))
+ return;
- if(! count($contacts)) {
- return;
+ // Count active workers and compare them with a maximum value that depends on the load
+ if (poller_too_much_workers())
+ return;
}
- foreach($contacts as $c) {
-
- $res = q("SELECT * FROM `contact` WHERE `id` = %d LIMIT 1",
- intval($c['id'])
- );
-
- if((! $res) || (! count($res)))
- continue;
+}
- foreach($res as $contact) {
+function poller_too_much_workers() {
- $xml = false;
+ $queues = get_config("system", "worker_queues");
- if($manual_id)
- $contact['last-update'] = '0000-00-00 00:00:00';
+ if ($queues == 0)
+ $queues = 4;
- if(in_array($contact['network'], array(NETWORK_DFRN, NETWORK_ZOT, NETWORK_OSTATUS)))
- $contact['priority'] = 2;
+ $active = poller_active_workers();
- if($contact['subhub'] AND in_array($contact['network'], array(NETWORK_DFRN, NETWORK_ZOT, NETWORK_OSTATUS))) {
- // 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'.
+ // Decrease the number of workers at higher load
+ if(function_exists('sys_getloadavg')) {
+ $load = max(sys_getloadavg());
- $poll_interval = get_config('system','pushpoll_frequency');
- $contact['priority'] = (($poll_interval !== false) ? intval($poll_interval) : 3);
- }
+ $maxsysload = intval(get_config('system','maxloadavg'));
+ if($maxsysload < 1)
+ $maxsysload = 50;
- if($contact['priority'] AND !$force) {
-
- $update = false;
-
- $t = $contact['last-update'];
-
- /**
- * 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)
- continue;
- }
+ $maxworkers = $queues;
- logger("Polling ".$contact["network"]." ".$contact["id"]." ".$contact["nick"]." ".$contact["name"]);
+ // Some magical mathemathics to reduce the workers
+ $exponent = 3;
+ $slope = $maxworkers / pow($maxsysload, $exponent);
+ $queues = ceil($slope * pow(max(0, $maxsysload - $load), $exponent));
- proc_run('php','include/onepoll.php',$contact['id']);
+ logger("Current load: ".$load." - maximum: ".$maxsysload." - current queues: ".$active." - maximum: ".$queues, LOGGER_DEBUG);
- if($interval)
- @time_sleep_until(microtime(true) + (float) $interval);
- }
}
- logger('poller: end');
+ return($active >= $queues);
+}
+
+function poller_active_workers() {
+ $workers = q("SELECT COUNT(*) AS `workers` FROM `workerqueue` WHERE `executed` != '0000-00-00 00:00:00'");
- return;
+ return($workers[0]["workers"]);
}
if (array_search(__file__,get_included_files())===0){
poller_run($_SERVER["argv"],$_SERVER["argc"]);
killme();
}
+?>