]> git.mxchange.org Git - friendica.git/blob - include/cron.php
The table optimisation level calculation is now using the index size as well
[friendica.git] / include / cron.php
1 <?php
2 if (!file_exists("boot.php") AND (sizeof($_SERVER["argv"]) != 0)) {
3         $directory = dirname($_SERVER["argv"][0]);
4
5         if (substr($directory, 0, 1) != "/")
6                 $directory = $_SERVER["PWD"]."/".$directory;
7
8         $directory = realpath($directory."/..");
9
10         chdir($directory);
11 }
12
13 require_once("boot.php");
14
15
16 function cron_run(&$argv, &$argc){
17         global $a, $db;
18
19         if(is_null($a)) {
20                 $a = new App;
21         }
22
23         if(is_null($db)) {
24                 @include(".htconfig.php");
25                 require_once("include/dba.php");
26                 $db = new dba($db_host, $db_user, $db_pass, $db_data);
27                 unset($db_host, $db_user, $db_pass, $db_data);
28         };
29
30
31         require_once('include/session.php');
32         require_once('include/datetime.php');
33         require_once('include/items.php');
34         require_once('include/Contact.php');
35         require_once('include/email.php');
36         require_once('include/socgraph.php');
37         require_once('mod/nodeinfo.php');
38
39         load_config('config');
40         load_config('system');
41
42         // Don't check this stuff if the function is called by the poller
43         if (App::callstack() != "poller_run") {
44                 if (App::maxload_reached())
45                         return;
46                 if (App::is_already_running('cron', 'include/cron.php', 540))
47                         return;
48         }
49
50         $last = get_config('system','last_cron');
51
52         $poll_interval = intval(get_config('system','cron_interval'));
53         if(! $poll_interval)
54                 $poll_interval = 10;
55
56         if($last) {
57                 $next = $last + ($poll_interval * 60);
58                 if($next > time()) {
59                         logger('cron intervall not reached');
60                         return;
61                 }
62         }
63
64         $a->set_baseurl(get_config('system','url'));
65
66         load_hooks();
67
68         logger('cron: start');
69
70         // run queue delivery process in the background
71
72         proc_run('php',"include/queue.php");
73
74         // run diaspora photo queue process in the background
75
76         proc_run('php',"include/dsprphotoq.php");
77
78         // run the process to discover global contacts in the background
79
80         proc_run('php',"include/discover_poco.php");
81
82         // run the process to update locally stored global contacts in the background
83
84         proc_run('php',"include/discover_poco.php", "checkcontact");
85
86         // expire any expired accounts
87
88         q("UPDATE user SET `account_expired` = 1 where `account_expired` = 0
89                 AND `account_expires_on` != '0000-00-00 00:00:00'
90                 AND `account_expires_on` < UTC_TIMESTAMP() ");
91
92         // delete user and contact records for recently removed accounts
93
94         $r = q("SELECT * FROM `user` WHERE `account_removed` = 1 AND `account_expires_on` < UTC_TIMESTAMP() - INTERVAL 3 DAY");
95         if ($r) {
96                 foreach($r as $user) {
97                         q("DELETE FROM `contact` WHERE `uid` = %d", intval($user['uid']));
98                         q("DELETE FROM `user` WHERE `uid` = %d", intval($user['uid']));
99                 }
100         }
101
102         $abandon_days = intval(get_config('system','account_abandon_days'));
103         if($abandon_days < 1)
104                 $abandon_days = 0;
105
106         // Check OStatus conversations
107         // Check only conversations with mentions (for a longer time)
108         check_conversations(true);
109
110         // Check every conversation
111         check_conversations(false);
112
113         // Set the gcontact-id in the item table if missing
114         item_set_gcontact();
115
116         // update nodeinfo data
117         nodeinfo_cron();
118
119         /// @TODO Regenerate usage statistics
120         // q("ANALYZE TABLE `item`");
121
122         // once daily run birthday_updates and then expire in background
123
124         $d1 = get_config('system','last_expire_day');
125         $d2 = intval(datetime_convert('UTC','UTC','now','d'));
126
127         if($d2 != intval($d1)) {
128
129                 update_contact_birthdays();
130
131                 proc_run('php',"include/discover_poco.php", "suggestions");
132
133                 set_config('system','last_expire_day',$d2);
134
135                 proc_run('php','include/expire.php');
136         }
137
138         // Clear cache entries
139         cron_clear_cache($a);
140
141         // Repair missing Diaspora values in contacts
142         cron_repair_diaspora($a);
143
144         // Repair entries in the database
145         cron_repair_database();
146
147         $manual_id  = 0;
148         $generation = 0;
149         $force      = false;
150         $restart    = false;
151
152         if(($argc > 1) && ($argv[1] == 'force'))
153                 $force = true;
154
155         if(($argc > 1) && ($argv[1] == 'restart')) {
156                 $restart = true;
157                 $generation = intval($argv[2]);
158                 if(! $generation)
159                         killme();
160         }
161
162         if(($argc > 1) && intval($argv[1])) {
163                 $manual_id = intval($argv[1]);
164                 $force     = true;
165         }
166
167         $interval = intval(get_config('system','poll_interval'));
168         if(! $interval)
169                 $interval = ((get_config('system','delivery_interval') === false) ? 3 : intval(get_config('system','delivery_interval')));
170
171         // If we are using the worker we don't need a delivery interval
172         if (get_config("system", "worker"))
173                 $interval = false;
174
175         $sql_extra = (($manual_id) ? " AND `id` = $manual_id " : "");
176
177         reload_plugins();
178
179         $d = datetime_convert();
180
181         // Only poll from those with suitable relationships,
182         // and which have a polling address and ignore Diaspora since
183         // we are unable to match those posts with a Diaspora GUID and prevent duplicates.
184
185         $abandon_sql = (($abandon_days)
186                 ? sprintf(" AND `user`.`login_date` > UTC_TIMESTAMP() - INTERVAL %d DAY ", intval($abandon_days))
187                 : ''
188         );
189
190         $contacts = q("SELECT `contact`.`id` FROM `contact` INNER JOIN `user` ON `user`.`uid` = `contact`.`uid`
191                 WHERE `rel` IN (%d, %d) AND `poll` != '' AND `network` IN ('%s', '%s', '%s', '%s', '%s', '%s')
192                 $sql_extra
193                 AND NOT `self` AND NOT `contact`.`blocked` AND NOT `contact`.`readonly` AND NOT `contact`.`archive`
194                 AND NOT `user`.`account_expired` AND NOT `user`.`account_removed` $abandon_sql ORDER BY RAND()",
195                 intval(CONTACT_IS_SHARING),
196                 intval(CONTACT_IS_FRIEND),
197                 dbesc(NETWORK_DFRN),
198                 dbesc(NETWORK_ZOT),
199                 dbesc(NETWORK_OSTATUS),
200                 dbesc(NETWORK_FEED),
201                 dbesc(NETWORK_MAIL),
202                 dbesc(NETWORK_MAIL2)
203         );
204
205         if(! count($contacts)) {
206                 return;
207         }
208
209         foreach($contacts as $c) {
210
211                 $res = q("SELECT * FROM `contact` WHERE `id` = %d LIMIT 1",
212                         intval($c['id'])
213                 );
214
215                 if((! $res) || (! count($res)))
216                         continue;
217
218                 foreach($res as $contact) {
219
220                         $xml = false;
221
222                         if($manual_id)
223                                 $contact['last-update'] = '0000-00-00 00:00:00';
224
225                         if(in_array($contact['network'], array(NETWORK_DFRN, NETWORK_ZOT, NETWORK_OSTATUS)))
226                                 $contact['priority'] = 2;
227
228                         if($contact['subhub'] AND in_array($contact['network'], array(NETWORK_DFRN, NETWORK_ZOT, NETWORK_OSTATUS))) {
229                                 // We should be getting everything via a hub. But just to be sure, let's check once a day.
230                                 // (You can make this more or less frequent if desired by setting 'pushpoll_frequency' appropriately)
231                                 // This also lets us update our subscription to the hub, and add or replace hubs in case it
232                                 // changed. We will only update hubs once a day, regardless of 'pushpoll_frequency'.
233
234                                 $poll_interval = get_config('system','pushpoll_frequency');
235                                 $contact['priority'] = (($poll_interval !== false) ? intval($poll_interval) : 3);
236                         }
237
238                         if($contact['priority'] AND !$force) {
239
240                                 $update     = false;
241
242                                 $t = $contact['last-update'];
243
244                                 /**
245                                  * Based on $contact['priority'], should we poll this site now? Or later?
246                                  */
247
248                                 switch ($contact['priority']) {
249                                         case 5:
250                                                 if(datetime_convert('UTC','UTC', 'now') > datetime_convert('UTC','UTC', $t . " + 1 month"))
251                                                         $update = true;
252                                                 break;
253                                         case 4:
254                                                 if(datetime_convert('UTC','UTC', 'now') > datetime_convert('UTC','UTC', $t . " + 1 week"))
255                                                         $update = true;
256                                                 break;
257                                         case 3:
258                                                 if(datetime_convert('UTC','UTC', 'now') > datetime_convert('UTC','UTC', $t . " + 1 day"))
259                                                         $update = true;
260                                                 break;
261                                         case 2:
262                                                 if(datetime_convert('UTC','UTC', 'now') > datetime_convert('UTC','UTC', $t . " + 12 hour"))
263                                                         $update = true;
264                                                 break;
265                                         case 1:
266                                         default:
267                                                 if(datetime_convert('UTC','UTC', 'now') > datetime_convert('UTC','UTC', $t . " + 1 hour"))
268                                                         $update = true;
269                                                 break;
270                                 }
271                                 if(!$update)
272                                         continue;
273                         }
274
275                         logger("Polling ".$contact["network"]." ".$contact["id"]." ".$contact["nick"]." ".$contact["name"]);
276
277                         proc_run('php','include/onepoll.php',$contact['id']);
278
279                         if($interval)
280                                 @time_sleep_until(microtime(true) + (float) $interval);
281                 }
282         }
283
284         logger('cron: end');
285
286         set_config('system','last_cron', time());
287
288         return;
289 }
290
291 /**
292  * @brief Clear cache entries
293  *
294  * @param App $a
295  */
296 function cron_clear_cache(&$a) {
297
298         $last = get_config('system','cache_last_cleared');
299
300         if($last) {
301                 $next = $last + (3600); // Once per hour
302                 $clear_cache = ($next <= time());
303         } else
304                 $clear_cache = true;
305
306         if (!$clear_cache)
307                 return;
308
309         // clear old cache
310         Cache::clear();
311
312         // clear old item cache files
313         clear_cache();
314
315         // clear cache for photos
316         clear_cache($a->get_basepath(), $a->get_basepath()."/photo");
317
318         // clear smarty cache
319         clear_cache($a->get_basepath()."/view/smarty3/compiled", $a->get_basepath()."/view/smarty3/compiled");
320
321         // clear cache for image proxy
322         if (!get_config("system", "proxy_disabled")) {
323                 clear_cache($a->get_basepath(), $a->get_basepath()."/proxy");
324
325                 $cachetime = get_config('system','proxy_cache_time');
326                 if (!$cachetime) $cachetime = PROXY_DEFAULT_TIME;
327
328                 q('DELETE FROM `photo` WHERE `uid` = 0 AND `resource-id` LIKE "pic:%%" AND `created` < NOW() - INTERVAL %d SECOND', $cachetime);
329         }
330
331         // Delete the cached OEmbed entries that are older than one year
332         q("DELETE FROM `oembed` WHERE `created` < NOW() - INTERVAL 1 YEAR");
333
334         // Delete the cached "parse_url" entries that are older than one year
335         q("DELETE FROM `parsed_url` WHERE `created` < NOW() - INTERVAL 1 YEAR");
336
337         // Maximum table size in megabyte
338         $max_tablesize = intval(get_config('system','optimize_max_tablesize')) * 1000000;
339         if ($max_tablesize == 0)
340                 $max_tablesize = 100 * 1000000; // Default are 100 MB
341
342         // Minimum fragmentation level in percent
343         $fragmentation_level = intval(get_config('system','optimize_fragmentation')) / 100;
344         if ($fragmentation_level == 0)
345                 $fragmentation_level = 0.3; // Default value is 30%
346
347         // Optimize some tables that need to be optimized
348         $r = q("SHOW TABLE STATUS");
349         foreach($r as $table) {
350
351                 // Don't optimize tables that are too large
352                 if ($table["Data_length"] > $max_tablesize)
353                         continue;
354
355                 // Don't optimize empty tables
356                 if ($table["Data_length"] == 0)
357                         continue;
358
359                 // Calculate fragmentation
360                 $fragmentation = $table["Data_free"] / ($table["Data_length"] + $table["Index_length"]);
361
362                 logger("Table ".$table["Name"]." - Fragmentation level: ".round($fragmentation * 100, 2), LOGGER_DEBUG);
363
364                 // Don't optimize tables that needn't to be optimized
365                 if ($fragmentation < $fragmentation_level)
366                         continue;
367
368                 // So optimize it
369                 logger("Optimize Table ".$table["Name"], LOGGER_DEBUG);
370                 q("OPTIMIZE TABLE `%s`", dbesc($table["Name"]));
371         }
372
373         set_config('system','cache_last_cleared', time());
374 }
375
376 /**
377  * @brief Repair missing values in Diaspora contacts
378  *
379  * @param App $a
380  */
381 function cron_repair_diaspora(&$a) {
382         $r = q("SELECT `id`, `url` FROM `contact`
383                 WHERE `network` = '%s' AND (`batch` = '' OR `notify` = '' OR `poll` = '' OR pubkey = '')
384                         ORDER BY RAND() LIMIT 50", dbesc(NETWORK_DIASPORA));
385         if ($r) {
386                 foreach ($r AS $contact) {
387                         if (poco_reachable($contact["url"])) {
388                                 $data = probe_url($contact["url"]);
389                                 if ($data["network"] == NETWORK_DIASPORA) {
390                                         logger("Repair contact ".$contact["id"]." ".$contact["url"], LOGGER_DEBUG);
391                                         q("UPDATE `contact` SET `batch` = '%s', `notify` = '%s', `poll` = '%s', pubkey = '%s' WHERE `id` = %d",
392                                                 dbesc($data["batch"]), dbesc($data["notify"]), dbesc($data["poll"]), dbesc($data["pubkey"]),
393                                                 intval($contact["id"]));
394                                 }
395                         }
396                 }
397         }
398 }
399
400 /**
401  * @brief Do some repairs in database entries
402  *
403  */
404 function cron_repair_database() {
405
406         // Set the parent if it wasn't set. (Shouldn't happen - but does sometimes)
407         // This call is very "cheap" so we can do it at any time without a problem
408         q("UPDATE `item` INNER JOIN `item` AS `parent` ON `parent`.`uri` = `item`.`parent-uri` AND `parent`.`uid` = `item`.`uid` SET `item`.`parent` = `parent`.`id` WHERE `item`.`parent` = 0");
409
410         /// @todo
411         /// - remove thread entries without item
412         /// - remove sign entries without item
413         /// - remove children when parent got lost
414         /// - set contact-id in item when not present
415 }
416
417 if (array_search(__file__,get_included_files())===0){
418         cron_run($_SERVER["argv"],$_SERVER["argc"]);
419         killme();
420 }