update_daemon2.php

   1 #!/usr/bin/env php
   2 <?php
   3         set_include_path(dirname(__FILE__) ."/include" . PATH_SEPARATOR .
   4                 get_include_path());
   5
   6         declare(ticks = 1);
   7         chdir(dirname(__FILE__));
   8
   9         define('DISABLE_SESSIONS', true);
  10
  11         require_once "version.php";
  12
  13         if (strpos(VERSION, ".99") !== false || getenv('DAEMON_XDEBUG')) {
  14                 define('DAEMON_EXTENDED_DEBUG', true);
  15         }
  16
  17         require_once "functions.php";
  18         require_once "rssfuncs.php";
  19         require_once "sanity_check.php";
  20         require_once "config.php";
  21         require_once "db.php";
  22         require_once "db-prefs.php";
  23
  24         // defaults
  25         define('PURGE_INTERVAL', 3600); // seconds
  26         define('MAX_CHILD_RUNTIME', 600); // seconds
  27         define('MAX_JOBS', 2);
  28         define('SPAWN_INTERVAL', DAEMON_SLEEP_INTERVAL); // seconds
  29
  30         if (!function_exists('pcntl_fork')) {
  31                 die("error: This script requires PHP compiled with PCNTL module.\n");
  32         }
  33
  34         $master_handlers_installed = false;
  35
  36         $children = array();
  37         $ctimes = array();
  38
  39         $last_checkpoint = -1;
  40
  41         function reap_children() {
  42                 global $children;
  43                 global $ctimes;
  44
  45                 $tmp = array();
  46
  47                 foreach ($children as $pid) {
  48                         if (pcntl_waitpid($pid, $status, WNOHANG) != $pid) {
  49
  50                                 if (file_is_locked("update_daemon-$pid.lock")) {
  51                                         array_push($tmp, $pid);
  52                                 } else {
  53                                         _debug("[reap_children] child $pid seems active but lockfile is unlocked.");
  54                                         unset($ctimes[$pid]);
  55
  56                                 }
  57                         } else {
  58                                 _debug("[reap_children] child $pid reaped.");
  59                                 unset($ctimes[$pid]);
  60                         }
  61                 }
  62
  63                 $children = $tmp;
  64
  65                 return count($tmp);
  66         }
  67
  68         function check_ctimes() {
  69                 global $ctimes;
  70
  71                 foreach (array_keys($ctimes) as $pid) {
  72                         $started = $ctimes[$pid];
  73
  74                         if (time() - $started > MAX_CHILD_RUNTIME) {
  75                                 _debug("[MASTER] child process $pid seems to be stuck, aborting...");
  76                                 posix_kill($pid, SIGKILL);
  77                         }
  78                 }
  79         }
  80
  81         function sigchld_handler($signal) {
  82                 $running_jobs = reap_children();
  83
  84                 _debug("[SIGCHLD] jobs left: $running_jobs");
  85
  86                 pcntl_waitpid(-1, $status, WNOHANG);
  87         }
  88
  89         function shutdown($caller_pid) {
  90                 if ($caller_pid == posix_getpid()) {
  91                         if (file_exists(LOCK_DIRECTORY . "/update_daemon.lock")) {
  92                                 _debug("removing lockfile (master)...");
  93                                 unlink(LOCK_DIRECTORY . "/update_daemon.lock");
  94                         }
  95                 }
  96         }
  97
  98         function task_shutdown() {
  99                 $pid = posix_getpid();
 100
 101                 if (file_exists(LOCK_DIRECTORY . "/update_daemon-$pid.lock")) {
 102                         _debug("removing lockfile ($pid)...");
 103                         unlink(LOCK_DIRECTORY . "/update_daemon-$pid.lock");
 104                 }
 105         }
 106
 107         function sigint_handler() {
 108                 _debug("[MASTER] SIG_INT received.\n");
 109                 shutdown(posix_getpid());
 110                 die;
 111         }
 112
 113         function task_sigint_handler() {
 114                 _debug("[TASK] SIG_INT received.\n");
 115                 task_shutdown();
 116                 die;
 117         }
 118
 119         pcntl_signal(SIGCHLD, 'sigchld_handler');
 120
 121         $longopts = array("log:",
 122                         "tasks:",
 123                         "interval:",
 124                         "quiet",
 125                         "help");
 126
 127         $options = getopt("", $longopts);
 128
 129         if (isset($options["help"]) ) {
 130                 print "Tiny Tiny RSS update daemon.\n\n";
 131                 print "Options:\n";
 132                 print "  --log FILE           - log messages to FILE\n";
 133                 print "  --tasks N            - amount of update tasks to spawn\n";
 134                 print "                         default: " . MAX_JOBS . "\n";
 135                 print "  --interval N         - task spawn interval\n";
 136                 print "                         default: " . SPAWN_INTERVAL . " seconds.\n";
 137                 print "  --quiet              - don't output messages to stdout\n";
 138                 return;
 139         }
 140
 141         define('QUIET', isset($options['quiet']));
 142
 143         if (isset($options["tasks"])) {
 144                 _debug("Set to spawn " . $options["tasks"] . " children.");
 145                 $max_jobs = $options["tasks"];
 146         } else {
 147                 $max_jobs = MAX_JOBS;
 148         }
 149
 150         if (isset($options["interval"])) {
 151                 _debug("Spawn interval: " . $options["interval"] . " seconds.");
 152                 $spawn_interval = $options["interval"];
 153         } else {
 154                 $spawn_interval = SPAWN_INTERVAL;
 155         }
 156
 157         if (isset($options["log"])) {
 158                 _debug("Logging to " . $options["log"]);
 159                 define('LOGFILE', $options["log"]);
 160         }
 161
 162         if (file_is_locked("update_daemon.lock")) {
 163                 die("error: Can't create lockfile. ".
 164                         "Maybe another daemon is already running.\n");
 165         }
 166
 167         // Try to lock a file in order to avoid concurrent update.
 168         $lock_handle = make_lockfile("update_daemon.lock");
 169
 170         if (!$lock_handle) {
 171                 die("error: Can't create lockfile. ".
 172                         "Maybe another daemon is already running.\n");
 173         }
 174
 175         // Testing database connection.
 176         // It is unnecessary to start the fork loop if database is not ok.
 177         $link = db_connect(DB_HOST, DB_USER, DB_PASS, DB_NAME);
 178
 179         if (!init_connection($link)) die("Can't initialize db connection.\n");
 180
 181         $schema_version = get_schema_version($link);
 182
 183         db_close($link);
 184
 185         while (true) {
 186
 187                 // Since sleep is interupted by SIGCHLD, we need another way to
 188                 // respect the spawn interval
 189                 $next_spawn = $last_checkpoint + $spawn_interval - time();
 190
 191                 if ($next_spawn % 60 == 0) {
 192                         $running_jobs = count($children);
 193                         _debug("[MASTER] active jobs: $running_jobs, next spawn at $next_spawn sec.");
 194                 }
 195
 196                 if ($last_checkpoint + $spawn_interval < time()) {
 197
 198                         /* Check if schema version changed */
 199
 200                         $link = db_connect(DB_HOST, DB_USER, DB_PASS, DB_NAME);
 201                         if (!init_connection($link)) die("Can't initialize db connection.\n");
 202                         $test_schema_version = get_schema_version($link);
 203                         db_close($link);
 204
 205                         if ($test_schema_version != $schema_version) {
 206                                 echo "Expected schema version: $schema_version, got: $test_schema_version\n";
 207                                 echo "Schema version changed while we were running, bailing out\n";
 208                                 exit(100);
 209                         }
 210
 211                         check_ctimes();
 212                         reap_children();
 213
 214                         for ($j = count($children); $j < $max_jobs; $j++) {
 215                                 $pid = pcntl_fork();
 216                                 if ($pid == -1) {
 217                                         die("fork failed!\n");
 218                                 } else if ($pid) {
 219
 220                                         if (!$master_handlers_installed) {
 221                                                 _debug("[MASTER] installing shutdown handlers");
 222                                                 pcntl_signal(SIGINT, 'sigint_handler');
 223                                                 register_shutdown_function('shutdown', posix_getpid());
 224                                                 $master_handlers_installed = true;
 225                                         }
 226
 227                                         _debug("[MASTER] spawned client $j [PID:$pid]...");
 228                                         array_push($children, $pid);
 229                                         $ctimes[$pid] = time();
 230                                 } else {
 231                                         pcntl_signal(SIGCHLD, SIG_IGN);
 232                                         pcntl_signal(SIGINT, 'task_sigint_handler');
 233
 234                                         register_shutdown_function('task_shutdown');
 235
 236                                         $my_pid = posix_getpid();
 237                                         $lock_filename = "update_daemon-$my_pid.lock";
 238
 239                                         $lock_handle = make_lockfile($lock_filename);
 240
 241                                         if (!$lock_handle) {
 242                                                 die("error: Can't create lockfile ($lock_filename). ".
 243                                                 "Maybe another daemon is already running.\n");
 244                                         }
 245
 246                                         // ****** Updating RSS code *******
 247                                         // Only run in fork process.
 248
 249                                         $start_timestamp = time();
 250
 251                                         $link = db_connect(DB_HOST, DB_USER, DB_PASS, DB_NAME);
 252
 253                                         if (!init_connection($link)) return;
 254
 255                                         // We disable stamp file, since it is of no use in a multiprocess update.
 256                                         // not really, tho for the time being -fox
 257                                         if (!make_stampfile('update_daemon.stamp')) {
 258                                                 _debug("warning: unable to create stampfile\n");
 259                                         }
 260
 261                                         // Call to the feed batch update function
 262                                         // and maybe regenerate feedbrowser cache
 263
 264                                         $nf = 0;
 265
 266                                         _debug("Waiting before update [$j]..");
 267                                         sleep($j*5);
 268                                         $nf = update_daemon_common($link);
 269
 270                                         if (rand(0,100) > 50) {
 271                                                 $count = update_feedbrowser_cache($link);
 272                                                 _debug("Feedbrowser updated, $count feeds processed.");
 273
 274                                                 purge_orphans($link, true);
 275
 276                                                 $rc = cleanup_tags($link, 14, 50000);
 277
 278                                                 _debug("Cleaned $rc cached tags.");
 279
 280                                                 global $pluginhost;
 281                                                 $pluginhost->run_hooks($pluginhost::HOOK_UPDATE_TASK, "hook_update_task", $op);
 282                                         }
 283
 284                                         _debug("Elapsed time: " . (time() - $start_timestamp) . " second(s)");
 285
 286                                         if ($nf > 0) {
 287                                                 _debug("Feeds processed: $nf");
 288
 289                                                 if (time() - $start_timestamp > 0) {
 290                                                         _debug("Feeds/minute: " . sprintf("%.2d", $nf/((time()-$start_timestamp)/60)));
 291                                                 }
 292                                         }
 293
 294                                         db_close($link);
 295
 296                                         // We are in a fork.
 297                                         // We wait a little before exiting to avoid to be faster than our parent process.
 298                                         sleep(1);
 299
 300                                         unlink(LOCK_DIRECTORY . "/$lock_filename");
 301
 302                                         // We exit in order to avoid fork bombing.
 303                                         exit(0);
 304                                 }
 305                         }
 306                         $last_checkpoint = time();
 307                 }
 308                 sleep(1);
 309         }
 310
 311 ?>