[tt-rss.git] / update_daemon2.php

#!/usr/bin/env php
<?php
	set_include_path(dirname(__FILE__) ."/include" . PATH_SEPARATOR .
		get_include_path());

	declare(ticks = 1);
	chdir(dirname(__FILE__));

	define('DISABLE_SESSIONS', true);

	require_once "version.php";

	if (strpos(VERSION, ".99") !== false || getenv('DAEMON_XDEBUG')) {
		define('DAEMON_EXTENDED_DEBUG', true);
	}

	require_once "autoload.php";
	require_once "functions.php";
	require_once "rssfuncs.php";
	require_once "sanity_check.php";
	require_once "config.php";
	require_once "db.php";
	require_once "db-prefs.php";
	require_once "errorhandler.php";

	// defaults
	define('PURGE_INTERVAL', 3600); // seconds
	define('MAX_CHILD_RUNTIME', 600); // seconds
	define('MAX_JOBS', 2);
	define('SPAWN_INTERVAL', DAEMON_SLEEP_INTERVAL); // seconds

	if (!function_exists('pcntl_fork')) {
		die("error: This script requires PHP compiled with PCNTL module.\n");
	}

	$master_handlers_installed = false;

	$children = array();
	$ctimes = array();

	$last_checkpoint = -1;

	function reap_children() {
		global $children;
		global $ctimes;

		$tmp = array();

		foreach ($children as $pid) {
			if (pcntl_waitpid($pid, $status, WNOHANG) != $pid) {

				if (file_is_locked("update_daemon-$pid.lock")) {
					array_push($tmp, $pid);
				} else {
					_debug("[reap_children] child $pid seems active but lockfile is unlocked.");
					unset($ctimes[$pid]);

				}
			} else {
				_debug("[reap_children] child $pid reaped.");
				unset($ctimes[$pid]);
			}
		}

		$children = $tmp;

		return count($tmp);
	}

	function check_ctimes() {
		global $ctimes;

		foreach (array_keys($ctimes) as $pid) {
			$started = $ctimes[$pid];

			if (time() - $started > MAX_CHILD_RUNTIME) {
				_debug("[MASTER] child process $pid seems to be stuck, aborting...");
				posix_kill($pid, SIGKILL);
			}
		}
	}

	function sigchld_handler($signal) {
		$running_jobs = reap_children();

		_debug("[SIGCHLD] jobs left: $running_jobs");

		pcntl_waitpid(-1, $status, WNOHANG);
	}

	function shutdown($caller_pid) {
		if ($caller_pid == posix_getpid()) {
			if (file_exists(LOCK_DIRECTORY . "/update_daemon.lock")) {
				_debug("removing lockfile (master)...");
				unlink(LOCK_DIRECTORY . "/update_daemon.lock");
			}
		}
	}

	function task_shutdown() {
		$pid = posix_getpid();

		if (file_exists(LOCK_DIRECTORY . "/update_daemon-$pid.lock")) {
			_debug("removing lockfile ($pid)...");
			unlink(LOCK_DIRECTORY . "/update_daemon-$pid.lock");
		}
	}

	function sigint_handler() {
		_debug("[MASTER] SIG_INT received.\n");
		shutdown(posix_getpid());
		die;
	}

	function task_sigint_handler() {
		_debug("[TASK] SIG_INT received.\n");
		task_shutdown();
		die;
	}

	pcntl_signal(SIGCHLD, 'sigchld_handler');

	$longopts = array("log:",
			"tasks:",
			"interval:",
			"quiet",
			"help");

	$options = getopt("", $longopts);

	if (isset($options["help"]) ) {
		print "Tiny Tiny RSS update daemon.\n\n";
		print "Options:\n";
		print "  --log FILE           - log messages to FILE\n";
		print "  --tasks N            - amount of update tasks to spawn\n";
		print "                         default: " . MAX_JOBS . "\n";
		print "  --interval N         - task spawn interval\n";
		print "                         default: " . SPAWN_INTERVAL . " seconds.\n";
		print "  --quiet              - don't output messages to stdout\n";
		return;
	}

	define('QUIET', isset($options['quiet']));

	if (isset($options["tasks"])) {
		_debug("Set to spawn " . $options["tasks"] . " children.");
		$max_jobs = $options["tasks"];
	} else {
		$max_jobs = MAX_JOBS;
	}

	if (isset($options["interval"])) {
		_debug("Spawn interval: " . $options["interval"] . " seconds.");
		$spawn_interval = $options["interval"];
	} else {
		$spawn_interval = SPAWN_INTERVAL;
	}

	if (isset($options["log"])) {
		_debug("Logging to " . $options["log"]);
		define('LOGFILE', $options["log"]);
	}

	if (file_is_locked("update_daemon.lock")) {
		die("error: Can't create lockfile. ".
			"Maybe another daemon is already running.\n");
	}

	// Try to lock a file in order to avoid concurrent update.
	$lock_handle = make_lockfile("update_daemon.lock");

	if (!$lock_handle) {
		die("error: Can't create lockfile. ".
			"Maybe another daemon is already running.\n");
	}

	// Testing database connection.
	// It is unnecessary to start the fork loop if database is not ok.
	$link = db_connect(DB_HOST, DB_USER, DB_PASS, DB_NAME);

	if (!init_plugins($link)) die("Can't initialize db connection.\n");

	$schema_version = get_schema_version($link);

	db_close($link);

	if ($schema_version != SCHEMA_VERSION) {
		die("Schema version is wrong, please upgrade the database.\n");
	}

	while (true) {

		// Since sleep is interupted by SIGCHLD, we need another way to
		// respect the spawn interval
		$next_spawn = $last_checkpoint + $spawn_interval - time();

		if ($next_spawn % 60 == 0) {
			$running_jobs = count($children);
			_debug("[MASTER] active jobs: $running_jobs, next spawn at $next_spawn sec.");
		}

		if ($last_checkpoint + $spawn_interval < time()) {

			/* Check if schema version changed */

			$link = db_connect(DB_HOST, DB_USER, DB_PASS, DB_NAME);
			if (!init_plugins($link)) die("Can't initialize db connection.\n");
			$test_schema_version = get_schema_version($link);
			db_close($link);

			if ($test_schema_version != $schema_version) {
				echo "Expected schema version: $schema_version, got: $test_schema_version\n";
				echo "Schema version changed while we were running, bailing out\n";
				exit(100);
			}

			check_ctimes();
			reap_children();

			for ($j = count($children); $j < $max_jobs; $j++) {
				$pid = pcntl_fork();
				if ($pid == -1) {
					die("fork failed!\n");
				} else if ($pid) {

					if (!$master_handlers_installed) {
						_debug("[MASTER] installing shutdown handlers");
						pcntl_signal(SIGINT, 'sigint_handler');
						register_shutdown_function('shutdown', posix_getpid());
						$master_handlers_installed = true;
					}

					_debug("[MASTER] spawned client $j [PID:$pid]...");
					array_push($children, $pid);
					$ctimes[$pid] = time();
				} else {
					pcntl_signal(SIGCHLD, SIG_IGN);
					pcntl_signal(SIGINT, 'task_sigint_handler');

					register_shutdown_function('task_shutdown');

					$my_pid = posix_getpid();
					$lock_filename = "update_daemon-$my_pid.lock";

					$lock_handle = make_lockfile($lock_filename);

					if (!$lock_handle) {
						die("error: Can't create lockfile ($lock_filename). ".
						"Maybe another daemon is already running.\n");
					}

					// ****** Updating RSS code *******
					// Only run in fork process.

					$start_timestamp = time();

					$link = db_connect(DB_HOST, DB_USER, DB_PASS, DB_NAME);

					if (!init_plugins($link)) return;

					// We disable stamp file, since it is of no use in a multiprocess update.
					// not really, tho for the time being -fox
					if (!make_stampfile('update_daemon.stamp')) {
						_debug("warning: unable to create stampfile\n");
					}

					// Call to the feed batch update function
					// and maybe regenerate feedbrowser cache

					$nf = 0;

					_debug("Waiting before update [$j]..");
					sleep($j*5);
					$nf = update_daemon_common($link);

					if (rand(0,100) > 50) {
						$count = update_feedbrowser_cache($link);
						_debug("Feedbrowser updated, $count feeds processed.");

						purge_orphans($link, true);

						$rc = cleanup_tags($link, 14, 50000);

						_debug("Cleaned $rc cached tags.");

						global $pluginhost;
						$pluginhost->run_hooks($pluginhost::HOOK_UPDATE_TASK, "hook_update_task", $op);
					}

					_debug("Elapsed time: " . (time() - $start_timestamp) . " second(s)");

					if ($nf > 0) {
						_debug("Feeds processed: $nf");

						if (time() - $start_timestamp > 0) {
							_debug("Feeds/minute: " . sprintf("%.2d", $nf/((time()-$start_timestamp)/60)));
						}
					}

					db_close($link);

					// We are in a fork.
					// We wait a little before exiting to avoid to be faster than our parent process.
					sleep(1);

					unlink(LOCK_DIRECTORY . "/$lock_filename");

					// We exit in order to avoid fork bombing.
					exit(0);
				}
			}
			$last_checkpoint = time();
		}
		sleep(1);
	}

?>
Commit	Line	Data
	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 "autoload.php";
	18	require_once "functions.php";
	19	require_once "rssfuncs.php";
	20	require_once "sanity_check.php";
	21	require_once "config.php";
	22	require_once "db.php";
	23	require_once "db-prefs.php";
	24	require_once "errorhandler.php";
	25
	26	// defaults
	27	define('PURGE_INTERVAL', 3600); // seconds
	28	define('MAX_CHILD_RUNTIME', 600); // seconds
	29	define('MAX_JOBS', 2);
	30	define('SPAWN_INTERVAL', DAEMON_SLEEP_INTERVAL); // seconds
	31
	32	if (!function_exists('pcntl_fork')) {
	33	die("error: This script requires PHP compiled with PCNTL module.\n");
	34	}
	35
	36	$master_handlers_installed = false;
	37
	38	$children = array();
	39	$ctimes = array();
	40
	41	$last_checkpoint = -1;
	42
	43	function reap_children() {
	44	global $children;
	45	global $ctimes;
	46
	47	$tmp = array();
	48
	49	foreach ($children as $pid) {
	50	if (pcntl_waitpid($pid, $status, WNOHANG) != $pid) {
	51
	52	if (file_is_locked("update_daemon-$pid.lock")) {
	53	array_push($tmp, $pid);
	54	} else {
	55	_debug("[reap_children] child $pid seems active but lockfile is unlocked.");
	56	unset($ctimes[$pid]);
	57
	58	}
	59	} else {
	60	_debug("[reap_children] child $pid reaped.");
	61	unset($ctimes[$pid]);
	62	}
	63	}
	64
	65	$children = $tmp;
	66
	67	return count($tmp);
	68	}
	69
	70	function check_ctimes() {
	71	global $ctimes;
	72
	73	foreach (array_keys($ctimes) as $pid) {
	74	$started = $ctimes[$pid];
	75
	76	if (time() - $started > MAX_CHILD_RUNTIME) {
	77	_debug("[MASTER] child process $pid seems to be stuck, aborting...");
	78	posix_kill($pid, SIGKILL);
	79	}
	80	}
	81	}
	82
	83	function sigchld_handler($signal) {
	84	$running_jobs = reap_children();
	85
	86	_debug("[SIGCHLD] jobs left: $running_jobs");
	87
	88	pcntl_waitpid(-1, $status, WNOHANG);
	89	}
	90
	91	function shutdown($caller_pid) {
	92	if ($caller_pid == posix_getpid()) {
	93	if (file_exists(LOCK_DIRECTORY . "/update_daemon.lock")) {
	94	_debug("removing lockfile (master)...");
	95	unlink(LOCK_DIRECTORY . "/update_daemon.lock");
	96	}
	97	}
	98	}
	99
	100	function task_shutdown() {
	101	$pid = posix_getpid();
	102
	103	if (file_exists(LOCK_DIRECTORY . "/update_daemon-$pid.lock")) {
	104	_debug("removing lockfile ($pid)...");
	105	unlink(LOCK_DIRECTORY . "/update_daemon-$pid.lock");
	106	}
	107	}
	108
	109	function sigint_handler() {
	110	_debug("[MASTER] SIG_INT received.\n");
	111	shutdown(posix_getpid());
	112	die;
	113	}
	114
	115	function task_sigint_handler() {
	116	_debug("[TASK] SIG_INT received.\n");
	117	task_shutdown();
	118	die;
	119	}
	120
	121	pcntl_signal(SIGCHLD, 'sigchld_handler');
	122
	123	$longopts = array("log:",
	124	"tasks:",
	125	"interval:",
	126	"quiet",
	127	"help");
	128
	129	$options = getopt("", $longopts);
	130
	131	if (isset($options["help"]) ) {
	132	print "Tiny Tiny RSS update daemon.\n\n";
	133	print "Options:\n";
	134	print " --log FILE - log messages to FILE\n";
	135	print " --tasks N - amount of update tasks to spawn\n";
	136	print " default: " . MAX_JOBS . "\n";
	137	print " --interval N - task spawn interval\n";
	138	print " default: " . SPAWN_INTERVAL . " seconds.\n";
	139	print " --quiet - don't output messages to stdout\n";
	140	return;
	141	}
	142
	143	define('QUIET', isset($options['quiet']));
	144
	145	if (isset($options["tasks"])) {
	146	_debug("Set to spawn " . $options["tasks"] . " children.");
	147	$max_jobs = $options["tasks"];
	148	} else {
	149	$max_jobs = MAX_JOBS;
	150	}
	151
	152	if (isset($options["interval"])) {
	153	_debug("Spawn interval: " . $options["interval"] . " seconds.");
	154	$spawn_interval = $options["interval"];
	155	} else {
	156	$spawn_interval = SPAWN_INTERVAL;
	157	}
	158
	159	if (isset($options["log"])) {
	160	_debug("Logging to " . $options["log"]);
	161	define('LOGFILE', $options["log"]);
	162	}
	163
	164	if (file_is_locked("update_daemon.lock")) {
	165	die("error: Can't create lockfile. ".
	166	"Maybe another daemon is already running.\n");
	167	}
	168
	169	// Try to lock a file in order to avoid concurrent update.
	170	$lock_handle = make_lockfile("update_daemon.lock");
	171
	172	if (!$lock_handle) {
	173	die("error: Can't create lockfile. ".
	174	"Maybe another daemon is already running.\n");
	175	}
	176
	177	// Testing database connection.
	178	// It is unnecessary to start the fork loop if database is not ok.
	179	$link = db_connect(DB_HOST, DB_USER, DB_PASS, DB_NAME);
	180
	181	if (!init_plugins($link)) die("Can't initialize db connection.\n");
	182
	183	$schema_version = get_schema_version($link);
	184
	185	db_close($link);
	186
	187	if ($schema_version != SCHEMA_VERSION) {
	188	die("Schema version is wrong, please upgrade the database.\n");
	189	}
	190
	191	while (true) {
	192
	193	// Since sleep is interupted by SIGCHLD, we need another way to
	194	// respect the spawn interval
	195	$next_spawn = $last_checkpoint + $spawn_interval - time();
	196
	197	if ($next_spawn % 60 == 0) {
	198	$running_jobs = count($children);
	199	_debug("[MASTER] active jobs: $running_jobs, next spawn at $next_spawn sec.");
	200	}
	201
	202	if ($last_checkpoint + $spawn_interval < time()) {
	203
	204	/* Check if schema version changed */
	205
	206	$link = db_connect(DB_HOST, DB_USER, DB_PASS, DB_NAME);
	207	if (!init_plugins($link)) die("Can't initialize db connection.\n");
	208	$test_schema_version = get_schema_version($link);
	209	db_close($link);
	210
	211	if ($test_schema_version != $schema_version) {
	212	echo "Expected schema version: $schema_version, got: $test_schema_version\n";
	213	echo "Schema version changed while we were running, bailing out\n";
	214	exit(100);
	215	}
	216
	217	check_ctimes();
	218	reap_children();
	219
	220	for ($j = count($children); $j < $max_jobs; $j++) {
	221	$pid = pcntl_fork();
	222	if ($pid == -1) {
	223	die("fork failed!\n");
	224	} else if ($pid) {
	225
	226	if (!$master_handlers_installed) {
	227	_debug("[MASTER] installing shutdown handlers");
	228	pcntl_signal(SIGINT, 'sigint_handler');
	229	register_shutdown_function('shutdown', posix_getpid());
	230	$master_handlers_installed = true;
	231	}
	232
	233	_debug("[MASTER] spawned client $j [PID:$pid]...");
	234	array_push($children, $pid);
	235	$ctimes[$pid] = time();
	236	} else {
	237	pcntl_signal(SIGCHLD, SIG_IGN);
	238	pcntl_signal(SIGINT, 'task_sigint_handler');
	239
	240	register_shutdown_function('task_shutdown');
	241
	242	$my_pid = posix_getpid();
	243	$lock_filename = "update_daemon-$my_pid.lock";
	244
	245	$lock_handle = make_lockfile($lock_filename);
	246
	247	if (!$lock_handle) {
	248	die("error: Can't create lockfile ($lock_filename). ".
	249	"Maybe another daemon is already running.\n");
	250	}
	251
	252	// **** Updating RSS code *****
	253	// Only run in fork process.
	254
	255	$start_timestamp = time();
	256
	257	$link = db_connect(DB_HOST, DB_USER, DB_PASS, DB_NAME);
	258
	259	if (!init_plugins($link)) return;
	260
	261	// We disable stamp file, since it is of no use in a multiprocess update.
	262	// not really, tho for the time being -fox
	263	if (!make_stampfile('update_daemon.stamp')) {
	264	_debug("warning: unable to create stampfile\n");
	265	}
	266
	267	// Call to the feed batch update function
	268	// and maybe regenerate feedbrowser cache
	269
	270	$nf = 0;
	271
	272	_debug("Waiting before update [$j]..");
	273	sleep($j*5);
	274	$nf = update_daemon_common($link);
	275
	276	if (rand(0,100) > 50) {
	277	$count = update_feedbrowser_cache($link);
	278	_debug("Feedbrowser updated, $count feeds processed.");
	279
	280	purge_orphans($link, true);
	281
	282	$rc = cleanup_tags($link, 14, 50000);
	283
	284	_debug("Cleaned $rc cached tags.");
	285
	286	global $pluginhost;
	287	$pluginhost->run_hooks($pluginhost::HOOK_UPDATE_TASK, "hook_update_task", $op);
	288	}
	289
	290	_debug("Elapsed time: " . (time() - $start_timestamp) . " second(s)");
	291
	292	if ($nf > 0) {
	293	_debug("Feeds processed: $nf");
	294
	295	if (time() - $start_timestamp > 0) {
	296	_debug("Feeds/minute: " . sprintf("%.2d", $nf/((time()-$start_timestamp)/60)));
	297	}
	298	}
	299
	300	db_close($link);
	301
	302	// We are in a fork.
	303	// We wait a little before exiting to avoid to be faster than our parent process.
	304	sleep(1);
	305
	306	unlink(LOCK_DIRECTORY . "/$lock_filename");
	307
	308	// We exit in order to avoid fork bombing.
	309	exit(0);
	310	}
	311	}
	312	$last_checkpoint = time();
	313	}
	314	sleep(1);
	315	}
	316
	317	?>