X-Git-Url: https://git.wh0rd.org/?a=blobdiff_plain;f=update_daemon2.php;h=6410709adf37882be1df4a7dc51bc663f4c13bd4;hb=51cc387347668e98b1b8e6c6053581a2affc5fa6;hp=326b69e49dce1905d22f9c98c7567bb994fd4e9a;hpb=5a613536c7859f108da8d1e58de2ca0189a80efb;p=tt-rss.git

diff --git a/update_daemon2.php b/update_daemon2.php
old mode 100644
new mode 100755
index 326b69e4..6410709a
--- a/update_daemon2.php
+++ b/update_daemon2.php
@@ -1,19 +1,16 @@
-#!/usr/bin/php
+#!/usr/bin/env php
 <?php
+	set_include_path(get_include_path() . PATH_SEPARATOR .
+		dirname(__FILE__) . "/include");
+
 	// This is an experimental multiprocess update daemon.
 	// Some configurable variable may be found below.
 
-	// define('DEFAULT_ERROR_LEVEL', E_ALL);
-	define('DEFAULT_ERROR_LEVEL', E_ERROR | E_WARNING | E_PARSE);
-
 	declare(ticks = 1);
+	chdir(dirname(__FILE__));
 
-	define('MAGPIE_CACHE_DIR', '/var/tmp/magpie-ttrss-cache-daemon');
-	define('SIMPLEPIE_CACHE_DIR',	'/var/tmp/simplepie-ttrss-cache-daemon');
 	define('DISABLE_SESSIONS', true);
 
-	define('MAX_JOBS', 2);
-
 	require_once "version.php";
 
 	if (strpos(VERSION, ".99") !== false || getenv('DAEMON_XDEBUG')) {
@@ -21,37 +18,47 @@
 	}
 
 	define('PURGE_INTERVAL', 3600); // seconds
+	define('MAX_CHILD_RUNTIME', 600); // seconds
 
+	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 "lib/magpierss/rss_fetch.inc";
 
+	define('MAX_JOBS', 2);
 	define('SPAWN_INTERVAL', DAEMON_SLEEP_INTERVAL);
 
-	if (!ENABLE_UPDATE_DAEMON) {
-		die("Please enable option ENABLE_UPDATE_DAEMON in config.php\n");
+	if (!function_exists('pcntl_fork')) {
+		die("error: This script requires PHP compiled with PCNTL module.\n");
 	}
-	
-	require_once "db.php";
-	require_once "db-prefs.php";
-	require_once "functions.php";
-	require_once "magpierss/rss_fetch.inc";
-
-	error_reporting(DEFAULT_ERROR_LEVEL);
 
 	$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) != $pid) {
-				array_push($tmp, $pid);
+
+				if (file_is_locked("update_daemon-$pid.lock")) {
+					array_push($tmp, $pid);
+				} else {
+					_debug("[reap_children] child $pid seems active but lockfile is unlocked.");
+					unset($ctimes[$pid]);
+
+				}
 			} else {
-				_debug("[SIGCHLD] child $pid reaped.");
+				_debug("[reap_children] child $pid reaped.");
+				unset($ctimes[$pid]);
 			}
 		}
 
@@ -60,8 +67,17 @@
 		return count($tmp);
 	}
 
-	function sigalrm_handler() {
-		die("received SIGALRM, hang in feed update?\n");
+	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) {
@@ -72,26 +88,40 @@
 		pcntl_waitpid(-1, $status, WNOHANG);
 	}
 
-	function sigint_handler() {
-		unlink(LOCK_DIRECTORY . "/update_daemon.lock");
-		die("Received SIGINT. Exiting.\n");
+	function shutdown() {
+		if (file_exists(LOCK_DIRECTORY . "/update_daemon.lock"))
+			unlink(LOCK_DIRECTORY . "/update_daemon.lock");
 	}
 
-	pcntl_signal(SIGALRM, 'sigalrm_handler');
-	pcntl_signal(SIGCHLD, 'sigchld_handler');
-	pcntl_signal(SIGINT, 'sigint_handler');
+	function task_shutdown() {
+		$pid = posix_getpid();
 
-	if (file_is_locked("update_daemon.lock")) {
-		die("error: Can't create lockfile. ".
-			"Maybe another daemon is already running.\n");
+		if (file_exists(LOCK_DIRECTORY . "/update_daemon-$pid.lock"))
+			unlink(LOCK_DIRECTORY . "/update_daemon-$pid.lock");
 	}
 
+	function sigint_handler() {
+		shutdown();
+		die("[SIGINT] removing lockfile and exiting.\n");
+	}
+
+	function task_sigint_handler() {
+		task_shutdown();
+		die("[SIGINT] removing lockfile and exiting.\n");
+	}
+
+	pcntl_signal(SIGCHLD, 'sigchld_handler');
+
 	if (file_is_locked("update_daemon.lock")) {
 		die("error: Can't create lockfile. ".
 			"Maybe another daemon is already running.\n");
 	}
 
 	if (!pcntl_fork()) {
+		pcntl_signal(SIGINT, 'sigint_handler');
+		register_shutdown_function('shutdown');
+
+		// Try to lock a file in order to avoid concurrent update.
 		$lock_handle = make_lockfile("update_daemon.lock");
 
 		if (!$lock_handle) {
@@ -104,21 +134,16 @@
 
 	// 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);	
+	$link = db_connect(DB_HOST, DB_USER, DB_PASS, DB_NAME);
 
-	if (!$link) {
-		if (DB_TYPE == "mysql") {
-			print mysql_error();
-		}
-		// PG seems to display its own errors just fine by default.		
-		return;
-	}
+	if (!init_connection($link)) return;
 
 	db_close($link);
 
-
 	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 % 10 == 0) {
@@ -128,6 +153,7 @@
 
 		if ($last_checkpoint + SPAWN_INTERVAL < time()) {
 
+			check_ctimes();
 			reap_children();
 
 			for ($j = count($children); $j < MAX_JOBS; $j++) {
@@ -137,161 +163,75 @@
 				} else if ($pid) {
 					_debug("[MASTER] spawned client $j [PID:$pid]...");
 					array_push($children, $pid);
+					$ctimes[$pid] = time();
 				} else {
 					pcntl_signal(SIGCHLD, SIG_IGN);
-					pcntl_signal(SIGINT, SIG_DFL);
+					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);	
+					$link = db_connect(DB_HOST, DB_USER, DB_PASS, DB_NAME);
 
-					if (!$link) {
-						if (DB_TYPE == "mysql") {
-							print mysql_error();
-						}
-						// PG seems to display its own errors just fine by default.		
-						return;
-					}
-
-					if (DB_TYPE == "pgsql") {
-						pg_query("set client_encoding = 'utf-8'");
-						pg_set_client_encoding("UNICODE");
-					} else {
-						if (defined('MYSQL_CHARSET') && MYSQL_CHARSET) {
-							db_query($link, "SET NAMES " . MYSQL_CHARSET);
-							// db_query($link, "SET CHARACTER SET " . MYSQL_CHARSET);
-						}
-					}
+					if (!init_connection($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')) {
 						print "warning: unable to create stampfile";
-					}	
-
-					// $last_purge = 0;
-
-					// if (time() - $last_purge > PURGE_INTERVAL) {
-
-					// FIXME : $last_purge is of no use in a multiprocess update.
-					// FIXME : We ALWAYS purge old posts.
-					_debug("Purging old posts (random 30 feeds)...");
-					global_purge_old_posts($link, true, 30);
-
-					// 	$last_purge = time();
-					// }
-
-					// Process all other feeds using last_updated and interval parameters
-
-					$random_qpart = sql_random_function();
-						
-					if (DAEMON_UPDATE_LOGIN_LIMIT > 0) {
-						if (DB_TYPE == "pgsql") {
-							$login_thresh_qpart = "AND ttrss_users.last_login >= NOW() - INTERVAL '".DAEMON_UPDATE_LOGIN_LIMIT." days'";
-						} else {
-							$login_thresh_qpart = "AND ttrss_users.last_login >= DATE_SUB(NOW(), INTERVAL ".DAEMON_UPDATE_LOGIN_LIMIT." DAY)";
-						}			
-					} else {
-						$login_thresh_qpart = "";
 					}
 
-					if (DB_TYPE == "pgsql") {
-						$update_limit_qpart = "AND ttrss_feeds.last_updated < NOW() - INTERVAL '".(DAEMON_SLEEP_INTERVAL*2)." seconds'";
-					} else {
-						$update_limit_qpart = "AND ttrss_feeds.last_updated < DATE_SUB(NOW(), INTERVAL ".(DAEMON_SLEEP_INTERVAL*2)." SECOND)";
-					}
-
-					if (DB_TYPE == "pgsql") {
-							$updstart_thresh_qpart = "AND (ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < NOW() - INTERVAL '120 seconds')";
-						} else {
-							$updstart_thresh_qpart = "AND (ttrss_feeds.last_update_started IS NULL OR ttrss_feeds.last_update_started < DATE_SUB(NOW(), INTERVAL 120 SECOND))";
-						}			
-
-					$result = db_query($link, "SELECT feed_url,ttrss_feeds.id,owner_uid,
-							SUBSTRING(last_updated,1,19) AS last_updated,
-							update_interval 
-						FROM 
-							ttrss_feeds,ttrss_users 
-						WHERE 
-							ttrss_users.id = owner_uid $login_thresh_qpart $update_limit_qpart 
-							$updstart_thresh_qpart
-						ORDER BY $random_qpart DESC LIMIT " . DAEMON_FEED_LIMIT);
-
-					$user_prefs_cache = array();
-
-					_debug(sprintf("Scheduled %d feeds to update...\n", db_num_rows($result)));
-
-					// Here is a little cache magic in order to minimize risk of double feed updates.
-					$feeds_to_update = array();
-					while ($line = db_fetch_assoc($result)) {
-						$feeds_to_update[$line['id']] = $line;
-					}
-
-					// We update the feed last update started date before anything else.
-					// There is no lag due to feed contents downloads
-					// It prevent an other process to update the same feed.
-					$feed_ids = array_keys($feeds_to_update);
-					if($feed_ids) {
-						db_query($link, sprintf("UPDATE ttrss_feeds SET last_update_started = NOW()
-							WHERE id IN (%s)", implode(',', $feed_ids)));
-					}
-
-					while ($line = array_pop($feeds_to_update)) {
-
-						$upd_intl = $line["update_interval"];
-						$user_id = $line["owner_uid"];
-
-						if (!$upd_intl || $upd_intl == 0) {
-							if (!$user_prefs_cache[$user_id]['DEFAULT_UPDATE_INTERVAL']) {			
-								$upd_intl = get_pref($link, 'DEFAULT_UPDATE_INTERVAL', $user_id);
-								$user_prefs_cache[$user_id]['DEFAULT_UPDATE_INTERVAL'] = $upd_intl;
-							} else {
-								$upd_intl = $user_prefs_cache[$user_id]['DEFAULT_UPDATE_INTERVAL'];
-							}
-						}
+					// Call to the feed batch update function
+					// or regenerate feedbrowser cache
 
-						if ($upd_intl < 0) { 
-				#				print "Updates disabled.\n";
-							continue; 
-						}
-
-						_debug("Feed: " . $line["feed_url"] . ", " . $line["last_updated"]);
-
-				//			_debug(sprintf("\tLU: %d, INTL: %d, UID: %d) ", 
-				//				time() - strtotime($line["last_updated"]), $upd_intl*60, $user_id));
-
-						if (!$line["last_updated"] || 
-							time() - strtotime($line["last_updated"]) > ($upd_intl * 60)) {
+					if (rand(0,100) > 30) {
+						update_daemon_common($link);
+					} else {
+						$count = update_feedbrowser_cache($link);
+						_debug("Feedbrowser updated, $count feeds processed.");
 
-							_debug("Updating...");
+						purge_orphans($link, true);
 
-							pcntl_alarm(300);
+						$rc = cleanup_tags($link, 14, 50000);
 
-							update_rss_feed($link, $line["feed_url"], $line["id"], true);	
+						_debug("Cleaned $rc cached tags.");
 
-							pcntl_alarm(0);
+						_debug("Updating linked feeds...");
+						get_linked_feeds($link);
 
-							sleep(1); // prevent flood (FIXME make this an option?)
-						} else {
-							_debug("Update not needed.");
-						}
 					}
 
-					if (DAEMON_SENDS_DIGESTS) send_headlines_digests($link);
-
-					print "Elapsed time: " . (time() - $start_timestamp) . " second(s)\n";
+					_debug("Elapsed time: " . (time() - $start_timestamp) . " second(s)");
 
 					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);
 				}
+
+				// We wait a little time before the next fork, in order to let the first fork
+				// mark the feeds it update :
+				sleep(1);
 			}
 			$last_checkpoint = time();
 		}