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         require_once "errorhandler.php";
  24
  25         // defaults
  26         define('PURGE_INTERVAL', 3600); // seconds
  27         define('MAX_CHILD_RUNTIME', 600); // seconds
  28         define('MAX_JOBS', 2);
  29         define('SPAWN_INTERVAL', DAEMON_SLEEP_INTERVAL); // seconds
  30
  31         if (!function_exists('pcntl_fork')) {
  32                 die("error: This script requires PHP compiled with PCNTL module.\n");
  33         }
  34
  35         $master_handlers_installed = false;
  36
  37         $children = array();
  38         $ctimes = array();
  39
  40         $last_checkpoint = -1;
  41
  42         function reap_children() {
  43                 global $children;
  44                 global $ctimes;
  45
  46                 $tmp = array();
  47
  48                 foreach ($children as $pid) {
  49                         if (pcntl_waitpid($pid, $status, WNOHANG) != $pid) {
  50
  51                                 if (file_is_locked("update_daemon-$pid.lock")) {
  52                                         array_push($tmp, $pid);
  53                                 } else {
  54                                         _debug("[reap_children] child $pid seems active but lockfile is unlocked.");
  55                                         unset($ctimes[$pid]);
  56
  57                                 }
  58                         } else {
  59                                 _debug("[reap_children] child $pid reaped.");
  60                                 unset($ctimes[$pid]);
  61                         }
  62                 }
  63
  64                 $children = $tmp;
  65
  66                 return count($tmp);
  67         }
  68
  69         function check_ctimes() {
  70                 global $ctimes;
  71
  72                 foreach (array_keys($ctimes) as $pid) {
  73                         $started = $ctimes[$pid];
  74
  75                         if (time() - $started > MAX_CHILD_RUNTIME) {
  76                                 _debug("[MASTER] child process $pid seems to be stuck, aborting...");
  77                                 posix_kill($pid, SIGKILL);
  78                         }
  79                 }
  80         }
  81
  82         function sigchld_handler($signal) {
  83                 $running_jobs = reap_children();
  84
  85                 _debug("[SIGCHLD] jobs left: $running_jobs");
  86
  87                 pcntl_waitpid(-1, $status, WNOHANG);
  88         }
  89
  90         function shutdown($caller_pid) {
  91                 if ($caller_pid == posix_getpid()) {
  92                         if (file_exists(LOCK_DIRECTORY . "/update_daemon.lock")) {
  93                                 _debug("removing lockfile (master)...");
  94                                 unlink(LOCK_DIRECTORY . "/update_daemon.lock");
  95                         }
  96                 }
  97         }
  98
  99         function task_shutdown() {
 100                 $pid = posix_getpid();
 101
 102                 if (file_exists(LOCK_DIRECTORY . "/update_daemon-$pid.lock")) {
 103                         _debug("removing lockfile ($pid)...");
 104                         unlink(LOCK_DIRECTORY . "/update_daemon-$pid.lock");
 105                 }
 106         }
 107
 108         function sigint_handler() {
 109                 _debug("[MASTER] SIG_INT received.\n");
 110                 shutdown(posix_getpid());
 111                 die;
 112         }
 113
 114         function task_sigint_handler() {
 115                 _debug("[TASK] SIG_INT received.\n");
 116                 task_shutdown();
 117                 die;
 118         }
 119
 120         pcntl_signal(SIGCHLD, 'sigchld_handler');
 121
 122         $longopts = array("log:",
 123                         "tasks:",
 124                         "interval:",
 125                         "quiet",
 126                         "help");
 127
 128         $options = getopt("", $longopts);
 129
 130         if (isset($options["help"]) ) {
 131                 print "Tiny Tiny RSS update daemon.\n\n";
 132                 print "Options:\n";
 133                 print "  --log FILE           - log messages to FILE\n";
 134                 print "  --tasks N            - amount of update tasks to spawn\n";
 135                 print "                         default: " . MAX_JOBS . "\n";
 136                 print "  --interval N         - task spawn interval\n";
 137                 print "                         default: " . SPAWN_INTERVAL . " seconds.\n";
 138                 print "  --quiet              - don't output messages to stdout\n";
 139                 return;
 140         }
 141
 142         define('QUIET', isset($options['quiet']));
 143
 144         if (isset($options["tasks"])) {
 145                 _debug("Set to spawn " . $options["tasks"] . " children.");
 146                 $max_jobs = $options["tasks"];
 147         } else {
 148                 $max_jobs = MAX_JOBS;
 149         }
 150
 151         if (isset($options["interval"])) {
 152                 _debug("Spawn interval: " . $options["interval"] . " seconds.");
 153                 $spawn_interval = $options["interval"];
 154         } else {
 155                 $spawn_interval = SPAWN_INTERVAL;
 156         }
 157
 158         if (isset($options["log"])) {
 159                 _debug("Logging to " . $options["log"]);
 160                 define('LOGFILE', $options["log"]);
 161         }
 162
 163         if (file_is_locked("update_daemon.lock")) {
 164                 die("error: Can't create lockfile. ".
 165                         "Maybe another daemon is already running.\n");
 166         }
 167
 168         // Try to lock a file in order to avoid concurrent update.
 169         $lock_handle = make_lockfile("update_daemon.lock");
 170
 171         if (!$lock_handle) {
 172                 die("error: Can't create lockfile. ".
 173                         "Maybe another daemon is already running.\n");
 174         }
 175
 176         // Testing database connection.
 177         // It is unnecessary to start the fork loop if database is not ok.
 178         $link = db_connect(DB_HOST, DB_USER, DB_PASS, DB_NAME);
 179
 180         if (!init_connection($link)) die("Can't initialize db connection.\n");
 181
 182         $schema_version = get_schema_version($link);
 183
 184         db_close($link);
 185
 186         if ($schema_version != SCHEMA_VERSION) {
 187                 die("Schema version is wrong, please upgrade the database.\n");
 188         }
 189
 190         while (true) {
 191
 192                 // Since sleep is interupted by SIGCHLD, we need another way to
 193                 // respect the spawn interval
 194                 $next_spawn = $last_checkpoint + $spawn_interval - time();
 195
 196                 if ($next_spawn % 60 == 0) {
 197                         $running_jobs = count($children);
 198                         _debug("[MASTER] active jobs: $running_jobs, next spawn at $next_spawn sec.");
 199                 }
 200
 201                 if ($last_checkpoint + $spawn_interval < time()) {
 202
 203                         /* Check if schema version changed */
 204
 205                         $link = db_connect(DB_HOST, DB_USER, DB_PASS, DB_NAME);
 206                         if (!init_connection($link)) die("Can't initialize db connection.\n");
 207                         $test_schema_version = get_schema_version($link);
 208                         db_close($link);
 209
 210                         if ($test_schema_version != $schema_version) {
 211                                 echo "Expected schema version: $schema_version, got: $test_schema_version\n";
 212                                 echo "Schema version changed while we were running, bailing out\n";
 213                                 exit(100);
 214                         }
 215
 216                         check_ctimes();
 217                         reap_children();
 218
 219                         for ($j = count($children); $j < $max_jobs; $j++) {
 220                                 $pid = pcntl_fork();
 221                                 if ($pid == -1) {
 222                                         die("fork failed!\n");
 223                                 } else if ($pid) {
 224
 225                                         if (!$master_handlers_installed) {
 226                                                 _debug("[MASTER] installing shutdown handlers");
 227                                                 pcntl_signal(SIGINT, 'sigint_handler');
 228                                                 register_shutdown_function('shutdown', posix_getpid());
 229                                                 $master_handlers_installed = true;
 230                                         }
 231
 232                                         _debug("[MASTER] spawned client $j [PID:$pid]...");
 233                                         array_push($children, $pid);
 234                                         $ctimes[$pid] = time();
 235                                 } else {
 236                                         pcntl_signal(SIGCHLD, SIG_IGN);
 237                                         pcntl_signal(SIGINT, 'task_sigint_handler');
 238
 239                                         register_shutdown_function('task_shutdown');
 240
 241                                         $my_pid = posix_getpid();
 242                                         $lock_filename = "update_daemon-$my_pid.lock";
 243
 244                                         $lock_handle = make_lockfile($lock_filename);
 245
 246                                         if (!$lock_handle) {
 247                                                 die("error: Can't create lockfile ($lock_filename). ".
 248                                                 "Maybe another daemon is already running.\n");
 249                                         }
 250
 251                                         // ****** Updating RSS code *******
 252                                         // Only run in fork process.
 253
 254                                         $start_timestamp = time();
 255
 256                                         $link = db_connect(DB_HOST, DB_USER, DB_PASS, DB_NAME);
 257
 258                                         if (!init_connection($link)) return;
 259
 260                                         // We disable stamp file, since it is of no use in a multiprocess update.
 261                                         // not really, tho for the time being -fox
 262                                         if (!make_stampfile('update_daemon.stamp')) {
 263                                                 _debug("warning: unable to create stampfile\n");
 264                                         }
 265
 266                                         // Call to the feed batch update function
 267                                         // and maybe regenerate feedbrowser cache
 268
 269                                         $nf = 0;
 270
 271                                         _debug("Waiting before update [$j]..");
 272                                         sleep($j*5);
 273                                         $nf = update_daemon_common($link);
 274
 275                                         if (rand(0,100) > 50) {
 276                                                 $count = update_feedbrowser_cache($link);
 277                                                 _debug("Feedbrowser updated, $count feeds processed.");
 278
 279                                                 purge_orphans($link, true);
 280
 281                                                 $rc = cleanup_tags($link, 14, 50000);
 282
 283                                                 _debug("Cleaned $rc cached tags.");
 284
 285                                                 global $pluginhost;
 286                                                 $pluginhost->run_hooks($pluginhost::HOOK_UPDATE_TASK, "hook_update_task", $op);
 287                                         }
 288
 289                                         _debug("Elapsed time: " . (time() - $start_timestamp) . " second(s)");
 290
 291                                         if ($nf > 0) {
 292                                                 _debug("Feeds processed: $nf");
 293
 294                                                 if (time() - $start_timestamp > 0) {
 295                                                         _debug("Feeds/minute: " . sprintf("%.2d", $nf/((time()-$start_timestamp)/60)));
 296                                                 }
 297                                         }
 298
 299                                         db_close($link);
 300
 301                                         // We are in a fork.
 302                                         // We wait a little before exiting to avoid to be faster than our parent process.
 303                                         sleep(1);
 304
 305                                         unlink(LOCK_DIRECTORY . "/$lock_filename");
 306
 307                                         // We exit in order to avoid fork bombing.
 308                                         exit(0);
 309                                 }
 310                         }
 311                         $last_checkpoint = time();
 312                 }
 313                 sleep(1);
 314         }
 315
 316 ?>