diff options
author | Brion Vibber <brion@pobox.com> | 2010-02-16 09:01:59 -0800 |
---|---|---|
committer | Brion Vibber <brion@pobox.com> | 2010-02-16 09:16:51 -0800 |
commit | c74aea589d5a79d7048470d44e457dffc8919ad3 (patch) | |
tree | 3b3bc865018857e5146e04c593ddb0f468669b0a /classes | |
parent | 3d0c3f0577fb1b0a83bb65ae6439f018932c5c38 (diff) |
Stomp queue restructuring for mass scalability:
- Multiplexing queues into groups and for multiple sites.
- Sharing vs breakout configurable per site and per queue via $config['queue']['breakout']
- Detect how many times a message is redelivered, discard if it's killed too many daemons
- count configurable with $config['queue']['max_retries']
- can dump the items to files in $config['queue']['dead_letter_dir']
Queue daemon memory & resource leak fixes:
- avoid unnecessary reconnections to memcached server (switch persistent connections back in on second initialization, assuming it's child process)
- monkey-patch for leaky .ini loads in DB_DataObject::databaseStructure() - was leaking 200k per active switch
- applied leak fixes to Status_network as well, using intermediate base Safe_DataObject for both it and Memcache_DataObject
Misc queue fixes:
- correct handling of child processes exiting due to signal termination instead of regular exit
- shutdown instead of infinite respawn loop if we're already past the soft memory limit at startup
- Added --all option for xmppdaemon... still opens one xmpp connection per site that has xmpp active
Cache updates:
- add Cache::increment() method with native support for memcached atomic increment
Diffstat (limited to 'classes')
-rw-r--r-- | classes/Memcached_DataObject.php | 52 | ||||
-rw-r--r-- | classes/Safe_DataObject.php | 250 | ||||
-rw-r--r-- | classes/Status_network.php | 16 |
3 files changed, 263 insertions, 55 deletions
diff --git a/classes/Memcached_DataObject.php b/classes/Memcached_DataObject.php index ab65c30ce..16c3d906c 100644 --- a/classes/Memcached_DataObject.php +++ b/classes/Memcached_DataObject.php @@ -19,58 +19,9 @@ if (!defined('STATUSNET') && !defined('LACONICA')) { exit(1); } -class Memcached_DataObject extends DB_DataObject +class Memcached_DataObject extends Safe_DataObject { /** - * Destructor to free global memory resources associated with - * this data object when it's unset or goes out of scope. - * DB_DataObject doesn't do this yet by itself. - */ - - function __destruct() - { - $this->free(); - if (method_exists('DB_DataObject', '__destruct')) { - parent::__destruct(); - } - } - - /** - * Magic function called at serialize() time. - * - * We use this to drop a couple process-specific references - * from DB_DataObject which can cause trouble in future - * processes. - * - * @return array of variable names to include in serialization. - */ - function __sleep() - { - $vars = array_keys(get_object_vars($this)); - $skip = array('_DB_resultid', '_link_loaded'); - return array_diff($vars, $skip); - } - - /** - * Magic function called at unserialize() time. - * - * Clean out some process-specific variables which might - * be floating around from a previous process's cached - * objects. - * - * Old cached objects may still have them. - */ - function __wakeup() - { - // Refers to global state info from a previous process. - // Clear this out so we don't accidentally break global - // state in *this* process. - $this->_DB_resultid = null; - // We don't have any local DBO refs, so clear these out. - $this->_link_loaded = false; - } - - /** * Wrapper for DB_DataObject's static lookup using memcached * as backing instead of an in-process cache array. * @@ -579,3 +530,4 @@ class Memcached_DataObject extends DB_DataObject return $c->set($cacheKey, $value); } } + diff --git a/classes/Safe_DataObject.php b/classes/Safe_DataObject.php new file mode 100644 index 000000000..021f7b506 --- /dev/null +++ b/classes/Safe_DataObject.php @@ -0,0 +1,250 @@ +<?php +/* + * StatusNet - the distributed open-source microblogging tool + * Copyright (C) 2010, StatusNet, Inc. + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU Affero General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU Affero General Public License for more details. + * + * You should have received a copy of the GNU Affero General Public License + * along with this program. If not, see <http://www.gnu.org/licenses/>. + */ + +if (!defined('STATUSNET') && !defined('LACONICA')) { exit(1); } + +/** + * Extended DB_DataObject to improve a few things: + * - free global resources from destructor + * - remove bogus global references from serialized objects + * - don't leak memory when loading already-used .ini files + * (eg when using the same schema on thousands of databases) + */ +class Safe_DataObject extends DB_DataObject +{ + /** + * Destructor to free global memory resources associated with + * this data object when it's unset or goes out of scope. + * DB_DataObject doesn't do this yet by itself. + */ + + function __destruct() + { + $this->free(); + if (method_exists('DB_DataObject', '__destruct')) { + parent::__destruct(); + } + } + + /** + * Magic function called at serialize() time. + * + * We use this to drop a couple process-specific references + * from DB_DataObject which can cause trouble in future + * processes. + * + * @return array of variable names to include in serialization. + */ + function __sleep() + { + $vars = array_keys(get_object_vars($this)); + $skip = array('_DB_resultid', '_link_loaded'); + return array_diff($vars, $skip); + } + + /** + * Magic function called at unserialize() time. + * + * Clean out some process-specific variables which might + * be floating around from a previous process's cached + * objects. + * + * Old cached objects may still have them. + */ + function __wakeup() + { + // Refers to global state info from a previous process. + // Clear this out so we don't accidentally break global + // state in *this* process. + $this->_DB_resultid = null; + // We don't have any local DBO refs, so clear these out. + $this->_link_loaded = false; + } + + + /** + * Work around memory-leak bugs... + * Had to copy-paste the whole function in order to patch a couple lines of it. + * Would be nice if this code was better factored. + * + * @param optional string name of database to assign / read + * @param optional array structure of database, and keys + * @param optional array table links + * + * @access public + * @return true or PEAR:error on wrong paramenters.. or false if no file exists.. + * or the array(tablename => array(column_name=>type)) if called with 1 argument.. (databasename) + */ + function databaseStructure() + { + + global $_DB_DATAOBJECT; + + // Assignment code + + if ($args = func_get_args()) { + + if (count($args) == 1) { + + // this returns all the tables and their structure.. + if (!empty($_DB_DATAOBJECT['CONFIG']['debug'])) { + $this->debug("Loading Generator as databaseStructure called with args",1); + } + + $x = new DB_DataObject; + $x->_database = $args[0]; + $this->_connect(); + $DB = &$_DB_DATAOBJECT['CONNECTIONS'][$this->_database_dsn_md5]; + + $tables = $DB->getListOf('tables'); + class_exists('DB_DataObject_Generator') ? '' : + require_once 'DB/DataObject/Generator.php'; + + foreach($tables as $table) { + $y = new DB_DataObject_Generator; + $y->fillTableSchema($x->_database,$table); + } + return $_DB_DATAOBJECT['INI'][$x->_database]; + } else { + + $_DB_DATAOBJECT['INI'][$args[0]] = isset($_DB_DATAOBJECT['INI'][$args[0]]) ? + $_DB_DATAOBJECT['INI'][$args[0]] + $args[1] : $args[1]; + + if (isset($args[1])) { + $_DB_DATAOBJECT['LINKS'][$args[0]] = isset($_DB_DATAOBJECT['LINKS'][$args[0]]) ? + $_DB_DATAOBJECT['LINKS'][$args[0]] + $args[2] : $args[2]; + } + return true; + } + + } + + + + if (!$this->_database) { + $this->_connect(); + } + + // loaded already? + if (!empty($_DB_DATAOBJECT['INI'][$this->_database])) { + + // database loaded - but this is table is not available.. + if ( + empty($_DB_DATAOBJECT['INI'][$this->_database][$this->__table]) + && !empty($_DB_DATAOBJECT['CONFIG']['proxy']) + ) { + if (!empty($_DB_DATAOBJECT['CONFIG']['debug'])) { + $this->debug("Loading Generator to fetch Schema",1); + } + class_exists('DB_DataObject_Generator') ? '' : + require_once 'DB/DataObject/Generator.php'; + + + $x = new DB_DataObject_Generator; + $x->fillTableSchema($this->_database,$this->__table); + } + return true; + } + + + if (empty($_DB_DATAOBJECT['CONFIG'])) { + DB_DataObject::_loadConfig(); + } + + // if you supply this with arguments, then it will take those + // as the database and links array... + + $schemas = isset($_DB_DATAOBJECT['CONFIG']['schema_location']) ? + array("{$_DB_DATAOBJECT['CONFIG']['schema_location']}/{$this->_database}.ini") : + array() ; + + if (isset($_DB_DATAOBJECT['CONFIG']["ini_{$this->_database}"])) { + $schemas = is_array($_DB_DATAOBJECT['CONFIG']["ini_{$this->_database}"]) ? + $_DB_DATAOBJECT['CONFIG']["ini_{$this->_database}"] : + explode(PATH_SEPARATOR,$_DB_DATAOBJECT['CONFIG']["ini_{$this->_database}"]); + } + + + /* BEGIN CHANGED FROM UPSTREAM */ + $_DB_DATAOBJECT['INI'][$this->_database] = $this->parseIniFiles($schemas); + /* END CHANGED FROM UPSTREAM */ + + // now have we loaded the structure.. + + if (!empty($_DB_DATAOBJECT['INI'][$this->_database][$this->__table])) { + return true; + } + // - if not try building it.. + if (!empty($_DB_DATAOBJECT['CONFIG']['proxy'])) { + class_exists('DB_DataObject_Generator') ? '' : + require_once 'DB/DataObject/Generator.php'; + + $x = new DB_DataObject_Generator; + $x->fillTableSchema($this->_database,$this->__table); + // should this fail!!!??? + return true; + } + $this->debug("Cant find database schema: {$this->_database}/{$this->__table} \n". + "in links file data: " . print_r($_DB_DATAOBJECT['INI'],true),"databaseStructure",5); + // we have to die here!! - it causes chaos if we dont (including looping forever!) + $this->raiseError( "Unable to load schema for database and table (turn debugging up to 5 for full error message)", DB_DATAOBJECT_ERROR_INVALIDARGS, PEAR_ERROR_DIE); + return false; + } + + /** For parseIniFiles */ + protected static $iniCache = array(); + + /** + * When switching site configurations, DB_DataObject was loading its + * .ini files over and over, leaking gobs of memory. + * This refactored helper function uses a local cache of .ini files + * to minimize the leaks. + * + * @param array of .ini file names $schemas + * @return array + */ + protected function parseIniFiles($schemas) + { + $key = implode("|", $schemas); + if (!isset(Safe_DataObject::$iniCache[$key])) { + $data = array(); + foreach ($schemas as $ini) { + if (file_exists($ini) && is_file($ini)) { + $data = array_merge($data, parse_ini_file($ini, true)); + + if (!empty($_DB_DATAOBJECT['CONFIG']['debug'])) { + if (!is_readable ($ini)) { + $this->debug("ini file is not readable: $ini","databaseStructure",1); + } else { + $this->debug("Loaded ini file: $ini","databaseStructure",1); + } + } + } else { + if (!empty($_DB_DATAOBJECT['CONFIG']['debug'])) { + $this->debug("Missing ini file: $ini","databaseStructure",1); + } + } + } + Safe_DataObject::$iniCache[$key] = $data; + } + + return Safe_DataObject::$iniCache[$key]; + } +} + diff --git a/classes/Status_network.php b/classes/Status_network.php index 4bda24b6a..a452c32ce 100644 --- a/classes/Status_network.php +++ b/classes/Status_network.php @@ -21,7 +21,7 @@ if (!defined('STATUSNET') && !defined('LACONICA')) { exit(1); } -class Status_network extends DB_DataObject +class Status_network extends Safe_DataObject { ###START_AUTOCODE /* the code below is auto generated do not remove the above tag */ @@ -57,6 +57,7 @@ class Status_network extends DB_DataObject ###END_AUTOCODE static $cache = null; + static $cacheInitialized = false; static $base = null; static $wildcard = null; @@ -78,11 +79,15 @@ class Status_network extends DB_DataObject if (class_exists('Memcache')) { self::$cache = new Memcache(); - // Can't close persistent connections, making forking painful. + // If we're a parent command-line process we need + // to be able to close out the connection after + // forking, so disable persistence. // - // @fixme only do this in *parent* CLI processes. - // single-process and child-processes *should* use persistent. - $persist = php_sapi_name() != 'cli'; + // We'll turn it back on again the second time + // through which will either be in a child process, + // or a single-process script which is switching + // configurations. + $persist = php_sapi_name() != 'cli' || self::$cacheInitialized; if (is_array($servers)) { foreach($servers as $server) { self::$cache->addServer($server, 11211, $persist); @@ -90,6 +95,7 @@ class Status_network extends DB_DataObject } else { self::$cache->addServer($servers, 11211, $persist); } + self::$cacheInitialized = true; } self::$base = $dbname; |