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