update_daemon2.php

   1 #!/usr/bin/php
   2 <?php
   3         // This is an experimental multiprocess update daemon.
   4         // Some configurable variable may be found below.
   5
   6         // define('DEFAULT_ERROR_LEVEL', E_ALL);
   7         define('DEFAULT_ERROR_LEVEL', E_ERROR | E_WARNING | E_PARSE);
   8
   9         declare(ticks = 1);
  10
  11         define('MAGPIE_CACHE_DIR', '/var/tmp/magpie-ttrss-cache-daemon');
  12         define('SIMPLEPIE_CACHE_DIR',   '/var/tmp/simplepie-ttrss-cache-daemon');
  13         define('DISABLE_SESSIONS', true);
  14
  15         define('MAX_JOBS', 2);
  16
  17         require_once "version.php";
  18
  19         if (strpos(VERSION, ".99") !== false || getenv('DAEMON_XDEBUG')) {
  20                 define('DAEMON_EXTENDED_DEBUG', true);
  21         }
  22
  23         define('PURGE_INTERVAL', 3600); // seconds
  24
  25         require_once "sanity_check.php";
  26         require_once "config.php";
  27
  28         define('SPAWN_INTERVAL', DAEMON_SLEEP_INTERVAL);
  29
  30         if (!ENABLE_UPDATE_DAEMON) {
  31                 die("Please enable option ENABLE_UPDATE_DAEMON in config.php\n");
  32         }
  33
  34         require_once "db.php";
  35         require_once "db-prefs.php";
  36         require_once "functions.php";
  37         require_once "magpierss/rss_fetch.inc";
  38
  39         error_reporting(DEFAULT_ERROR_LEVEL);
  40
  41         $children = array();
  42
  43         $last_checkpoint = -1;
  44
  45         function reap_children() {
  46                 global $children;
  47
  48                 $tmp = array();
  49
  50                 foreach ($children as $pid) {
  51                         if (pcntl_waitpid($pid, $status, WNOHANG) != $pid) {
  52                                 array_push($tmp, $pid);
  53                         } else {
  54                                 _debug("[SIGCHLD] child $pid reaped.");
  55                         }
  56                 }
  57
  58                 $children = $tmp;
  59
  60                 return count($tmp);
  61         }
  62
  63         function sigalrm_handler() {
  64                 die("received SIGALRM, hang in feed update?\n");
  65         }
  66
  67         function sigchld_handler($signal) {
  68                 $running_jobs = reap_children();
  69
  70                 _debug("[SIGCHLD] jobs left: $running_jobs");
  71
  72                 pcntl_waitpid(-1, $status, WNOHANG);
  73         }
  74
  75         function sigint_handler() {
  76                 unlink(LOCK_DIRECTORY . "/update_daemon.lock");
  77                 die("Received SIGINT. Exiting.\n");
  78         }
  79
  80         pcntl_signal(SIGALRM, 'sigalrm_handler');
  81         pcntl_signal(SIGCHLD, 'sigchld_handler');
  82         pcntl_signal(SIGINT, 'sigint_handler');
  83
  84         if (file_is_locked("update_daemon.lock")) {
  85                 die("error: Can't create lockfile. ".
  86                         "Maybe another daemon is already running.\n");
  87         }
  88
  89         if (file_is_locked("update_daemon.lock")) {
  90                 die("error: Can't create lockfile. ".
  91                         "Maybe another daemon is already running.\n");
  92         }
  93
  94         if (!pcntl_fork()) {
  95                 $lock_handle = make_lockfile("update_daemon.lock");
  96
  97                 if (!$lock_handle) {
  98                         die("error: Can't create lockfile. ".
  99                                 "Maybe another daemon is already running.\n");
 100                 }
 101
 102                 while (true) { sleep(100); }
 103         }
 104
 105         // Testing database connection.
 106         // It is unnecessary to start the fork loop if database is not ok.
 107         $link = db_connect(DB_HOST, DB_USER, DB_PASS, DB_NAME);
 108
 109         if (!$link) {
 110                 if (DB_TYPE == "mysql") {
 111                         print mysql_error();
 112                 }
 113                 // PG seems to display its own errors just fine by default.
 114                 return;
 115         }
 116
 117         db_close($link);
 118
 119
 120         while (true) {
 121
 122                 $next_spawn = $last_checkpoint + SPAWN_INTERVAL - time();
 123
 124                 if ($next_spawn % 10 == 0) {
 125                         $running_jobs = count($children);
 126                         _debug("[MASTER] active jobs: $running_jobs, next spawn at $next_spawn sec.");
 127                 }
 128
 129                 if ($last_checkpoint + SPAWN_INTERVAL < time()) {
 130
 131                         reap_children();
 132
 133                         for ($j = count($children); $j < MAX_JOBS; $j++) {
 134                                 $pid = pcntl_fork();
 135                                 if ($pid == -1) {
 136                                         die("fork failed!\n");
 137                                 } else if ($pid) {
 138                                         _debug("[MASTER] spawned client $j [PID:$pid]...");
 139                                         array_push($children, $pid);
 140                                 } else {
 141                                         pcntl_signal(SIGCHLD, SIG_IGN);
 142                                         pcntl_signal(SIGINT, SIG_DFL);
 143
 144                                         // ****** Updating RSS code *******
 145                                         // Only run in fork process.
 146
 147                                         $start_timestamp = time();
 148
 149                                         $link = db_connect(DB_HOST, DB_USER, DB_PASS, DB_NAME);
 150
 151                                         if (!$link) {
 152                                                 if (DB_TYPE == "mysql") {
 153                                                         print mysql_error();
 154                                                 }
 155                                                 // PG seems to display its own errors just fine by default.
 156                                                 return;
 157                                         }
 158
 159                                         if (DB_TYPE == "pgsql") {
 160                                                 pg_query("set client_encoding = 'utf-8'");
 161                                                 pg_set_client_encoding("UNICODE");
 162                                         } else {
 163                                                 if (defined('MYSQL_CHARSET') && MYSQL_CHARSET) {
 164                                                         db_query($link, "SET NAMES " . MYSQL_CHARSET);
 165                                                         // db_query($link, "SET CHARACTER SET " . MYSQL_CHARSET);
 166                                                 }
 167                                         }
 168
 169                                         // We disable stamp file, since it is of no use in a multiprocess update.
 170                                         // not really, tho for the time being -fox
 171                                         if (!make_stampfile('update_daemon.stamp')) {
 172                                                 print "warning: unable to create stampfile";
 173                                         }
 174
 175                                         // $last_purge = 0;
 176
 177                                         // if (time() - $last_purge > PURGE_INTERVAL) {
 178
 179                                         // FIXME : $last_purge is of no use in a multiprocess update.
 180                                         // FIXME : We ALWAYS purge old posts.
 181                                         _debug("Purging old posts (random 30 feeds)...");
 182                                         global_purge_old_posts($link, true, 30);
 183
 184                                         //      $last_purge = time();
 185                                         // }
 186
 187                                         // Process all other feeds using last_updated and interval parameters
 188
 189                                         $random_qpart = sql_random_function();
 190
 191                                         if (DAEMON_UPDATE_LOGIN_LIMIT > 0) {
 192                                                 if (DB_TYPE == "pgsql") {
 193                                                         $login_thresh_qpart = "AND ttrss_users.last_login >= NOW() - INTERVAL '".DAEMON_UPDATE_LOGIN_LIMIT." days'";
 194                                                 } else {
 195                                                         $login_thresh_qpart = "AND ttrss_users.last_login >= DATE_SUB(NOW(), INTERVAL ".DAEMON_UPDATE_LOGIN_LIMIT." DAY)";
 196                                                 }
 197                                         } else {
 198                                                 $login_thresh_qpart = "";
 199                                         }
 200
 201                                         if (DB_TYPE == "pgsql") {
 202                                                 $update_limit_qpart = "AND ttrss_feeds.last_updated < NOW() - INTERVAL '".(DAEMON_SLEEP_INTERVAL*2)." seconds'";
 203                                         } else {
 204                                                 $update_limit_qpart = "AND ttrss_feeds.last_updated < DATE_SUB(NOW(), INTERVAL ".(DAEMON_SLEEP_INTERVAL*2)." SECOND)";
 205                                         }
 206
 207                                         if (DB_TYPE == "pgsql") {
 208                                                         $updstart_thresh_qpart = "AND (ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < NOW() - INTERVAL '120 seconds')";
 209                                                 } else {
 210                                                         $updstart_thresh_qpart = "AND (ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < DATE_SUB(NOW(), INTERVAL 120 SECOND))";
 211                                                 }
 212
 213                                         $result = db_query($link, "SELECT feed_url,ttrss_feeds.id,owner_uid,
 214                                                         SUBSTRING(last_updated,1,19) AS last_updated,
 215                                                         update_interval
 216                                                 FROM
 217                                                         ttrss_feeds,ttrss_users
 218                                                 WHERE
 219                                                         ttrss_users.id = owner_uid $login_thresh_qpart $update_limit_qpart
 220                                                         $updstart_thresh_qpart
 221                                                 ORDER BY $random_qpart DESC LIMIT " . DAEMON_FEED_LIMIT);
 222
 223                                         $user_prefs_cache = array();
 224
 225                                         _debug(sprintf("Scheduled %d feeds to update...\n", db_num_rows($result)));
 226
 227                                         // Here is a little cache magic in order to minimize risk of double feed updates.
 228                                         $feeds_to_update = array();
 229                                         while ($line = db_fetch_assoc($result)) {
 230                                                 $feeds_to_update[$line['id']] = $line;
 231                                         }
 232
 233                                         // We update the feed last update started date before anything else.
 234                                         // There is no lag due to feed contents downloads
 235                                         // It prevent an other process to update the same feed.
 236                                         $feed_ids = array_keys($feeds_to_update);
 237                                         if($feed_ids) {
 238                                                 db_query($link, sprintf("UPDATE ttrss_feeds SET last_update_started = NOW()
 239                                                         WHERE id IN (%s)", implode(',', $feed_ids)));
 240                                         }
 241
 242                                         while ($line = array_pop($feeds_to_update)) {
 243
 244                                                 $upd_intl = $line["update_interval"];
 245                                                 $user_id = $line["owner_uid"];
 246
 247                                                 if (!$upd_intl || $upd_intl == 0) {
 248                                                         if (!$user_prefs_cache[$user_id]['DEFAULT_UPDATE_INTERVAL']) {
 249                                                                 $upd_intl = get_pref($link, 'DEFAULT_UPDATE_INTERVAL', $user_id);
 250                                                                 $user_prefs_cache[$user_id]['DEFAULT_UPDATE_INTERVAL'] = $upd_intl;
 251                                                         } else {
 252                                                                 $upd_intl = $user_prefs_cache[$user_id]['DEFAULT_UPDATE_INTERVAL'];
 253                                                         }
 254                                                 }
 255
 256                                                 if ($upd_intl < 0) {
 257                                 #                               print "Updates disabled.\n";
 258                                                         continue;
 259                                                 }
 260
 261                                                 _debug("Feed: " . $line["feed_url"] . ", " . $line["last_updated"]);
 262
 263                                 //                      _debug(sprintf("\tLU: %d, INTL: %d, UID: %d) ",
 264                                 //                              time() - strtotime($line["last_updated"]), $upd_intl*60, $user_id));
 265
 266                                                 if (!$line["last_updated"] ||
 267                                                         time() - strtotime($line["last_updated"]) > ($upd_intl * 60)) {
 268
 269                                                         _debug("Updating...");
 270
 271                                                         pcntl_alarm(300);
 272
 273                                                         update_rss_feed($link, $line["feed_url"], $line["id"], true);
 274
 275                                                         pcntl_alarm(0);
 276
 277                                                         sleep(1); // prevent flood (FIXME make this an option?)
 278                                                 } else {
 279                                                         _debug("Update not needed.");
 280                                                 }
 281                                         }
 282
 283                                         if (DAEMON_SENDS_DIGESTS) send_headlines_digests($link);
 284
 285                                         print "Elapsed time: " . (time() - $start_timestamp) . " second(s)\n";
 286
 287                                         db_close($link);
 288
 289                                         // We are in a fork.
 290                                         // We wait a little before exiting to avoid to be faster than our parent process.
 291                                         sleep(1);
 292                                         // We exit in order to avoid fork bombing.
 293                                         exit(0);
 294                                 }
 295                         }
 296                         $last_checkpoint = time();
 297                 }
 298                 sleep(1);
 299         }
 300
 301 ?>