From ffa7cbae80f82885ef539c3ef1e2b0b6c1534450 Mon Sep 17 00:00:00 2001 From: Andrew Dolgov Date: Wed, 23 Jan 2008 17:19:32 +0100 Subject: [PATCH] update_daemon2: import single-file version from landure --- update_daemon2.php | 215 ++++++++++++++++++++++++++++++++++++-- update_daemon2_client.php | 164 ----------------------------- 2 files changed, 206 insertions(+), 173 deletions(-) delete mode 100644 update_daemon2_client.php diff --git a/update_daemon2.php b/update_daemon2.php index d422c536..a03a2a79 100644 --- a/update_daemon2.php +++ b/update_daemon2.php @@ -1,23 +1,50 @@ #!/usr/bin/php 0) $running_jobs--; @@ -30,6 +57,7 @@ die("Received SIGINT. Exiting.\n"); } + pcntl_signal(SIGALRM, 'sigalrm_handler'); pcntl_signal(SIGCHLD, 'sigchld_handler'); pcntl_signal(SIGINT, 'sigint_handler'); @@ -38,6 +66,11 @@ "Maybe another daemon is already running.\n"); } + if (file_is_locked("update_daemon.lock")) { + die("error: Can't create lockfile. ". + "Maybe another daemon is already running.\n"); + } + if (!pcntl_fork()) { $lock_handle = make_lockfile("update_daemon.lock"); @@ -49,6 +82,21 @@ while (true) { sleep(100); } } + // Testing database connection. + // It is unnecessary to start the fork loop if database is not ok. + $link = db_connect(DB_HOST, DB_USER, DB_PASS, DB_NAME); + + if (!$link) { + if (DB_TYPE == "mysql") { + print mysql_error(); + } + // PG seems to display its own errors just fine by default. + return; + } + + db_close($link); + + while (true) { $next_spawn = $last_checkpoint + SPAWN_INTERVAL - time(); @@ -70,7 +118,156 @@ } else { pcntl_signal(SIGCHLD, SIG_IGN); pcntl_signal(SIGINT, SIG_DFL); - passthru(PHP_EXECUTABLE . ' ' . CLIENT_PROCESS); + + // ****** Updating RSS code ******* + // Only run in fork process. + + $start_timestamp = time(); + + $link = db_connect(DB_HOST, DB_USER, DB_PASS, DB_NAME); + + if (!$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); + } + } + + // 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']; + } + } + + 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)) { + + _debug("Updating..."); + + pcntl_alarm(300); + + update_rss_feed($link, $line["feed_url"], $line["id"], true); + + pcntl_alarm(0); + + sleep(1); // prevent flood (FIXME make this an option?) + } else { + _debug("Update not needed."); + } + } + + if (DAEMON_SENDS_DIGESTS) send_headlines_digests($link); + + print "Elapsed time: " . (time() - $start_timestamp) . " second(s)\n"; + + db_close($link); + + // We are in a fork. + // We wait a little before exiting to avoid to be faster than our parent process. + sleep(1); + // We exit in order to avoid fork bombing. exit(0); } } diff --git a/update_daemon2_client.php b/update_daemon2_client.php deleted file mode 100644 index 8934cbe4..00000000 --- a/update_daemon2_client.php +++ /dev/null @@ -1,164 +0,0 @@ -#!/usr/bin/php - PURGE_INTERVAL) { - _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))); - - while ($line = db_fetch_assoc($result)) { - - $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']; - } - } - - 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)) { - - _debug("Updating..."); - - pcntl_alarm(300); - - update_rss_feed($link, $line["feed_url"], $line["id"], true); - - pcntl_alarm(0); - - sleep(1); // prevent flood (FIXME make this an option?) - } else { - _debug("Update not needed."); - } - } - - if (DAEMON_SENDS_DIGESTS) send_headlines_digests($link); - - print "Elapsed time: " . (time() - $start_timestamp) . " second(s)\n"; - - db_close($link); -?> -- 2.39.5