* Correct blob caching to reduce redundant blob loads on backups

diff --git a/RELEASE-NOTES b/RELEASE-NOTES
index 5f15b0b..6541a09 100644 (file)
--- a/RELEASE-NOTES
+++ b/RELEASE-NOTES
@@ -109,6 +109,7 @@ fully support the editing toolbar, but was found to be too confusing.
 * (bug 1423) LanguageJa.php update
 * Clean up duplicate code for selection of changeslist style
 * (bug 3475) anon contrib links on Special:Newpages
+* Correct blob caching to reduce redundant blob loads on backups
 
 
 === Caveats ===

diff --git a/includes/ExternalStoreDB.php b/includes/ExternalStoreDB.php
index 3c61726..d07e23b 100644 (file)
--- a/includes/ExternalStoreDB.php
+++ b/includes/ExternalStoreDB.php
@@ -11,20 +11,38 @@ require_once( 'LoadBalancer.php' );
 
 /** @package MediaWiki */
 
+/**
+ * External database storage will use one (or more) separate connection pools
+ * from what the main wiki uses. If we load many revisions, such as when doing
+ * bulk backups or maintenance, we want to keep them around over the lifetime
+ * of the script.
+ *
+ * Associative array of LoadBalancer objects, indexed by cluster name.
+ */
+global $wgExternalLoadBalancers;
+$wgExternalLoadBalancers = array();
+
+/**
+ * One-step cache variable to hold base blobs; operations that
+ * pull multiple revisions may often pull multiple times from
+ * the same blob. By keeping the last-used one open, we avoid
+ * redundant unserialization and decompression overhead.
+ */
+global $wgExternalBlobCache;
+$wgExternalBlobCache = array();
+
 class ExternalStoreDB {
-       var $loadBalancers = array();
-       
        /**
         * Fetch data from given URL
         * @param string $url An url
         */
 
        function &getLoadBalancer( $cluster ) {
-               global $wgExternalServers;
-               if ( !array_key_exists( $cluster, $this->loadBalancers ) ) {
-                       $this->loadBalancers[$cluster] = LoadBalancer::newFromParams( $wgExternalServers[$cluster] );
+               global $wgExternalServers, $wgExternalLoadBalancers;
+               if ( !array_key_exists( $cluster, $wgExternalLoadBalancers ) ) {
+                       $wgExternalLoadBalancers[$cluster] = LoadBalancer::newFromParams( $wgExternalServers[$cluster] );
                }
-               return $this->loadBalancers[$cluster];
+               return $wgExternalLoadBalancers[$cluster];
        }
        
        function &getSlave( $cluster ) {
@@ -51,14 +69,39 @@ class ExternalStoreDB {
                        $itemID = false;
                }
 
-               $dbr =& $this->getSlave( $cluster );
-               $ret = $dbr->selectField( 'blobs', 'blob_text', array( 'blob_id' => $id ) ); 
+               $ret =& $this->fetchBlob( $cluster, $id, $itemID );
 
                if ( $itemID !== false ) {
-                       # Unserialise object and get item
-                       $obj = unserialize( $ret );
-                       $ret = $obj->getItem( $itemID );
+                       return $ret->getItem( $itemID );
+               }
+               return $ret;
+       }
+       
+       /**
+        * Fetch a blob item out of the database; a cache of the last-loaded
+        * blob will be kept so that multiple loads out of a multi-item blob
+        * can avoid redundant database access and decompression.
+        * @return mixed
+        * @access private
+        */
+       function &fetchBlob( $cluster, $id, $itemID ) {
+               global $wgExternalBlobCache;
+               $cacheID = ( $itemID === false ) ? "$cluster/$id" : "$cluster/$id/";
+               if( isset( $wgExternalBlobCache[$cacheID] ) ) {
+                       wfDebug( "ExternalStoreDB::fetchBlob cache hit on $cacheID\n" );
+                       return $wgExternalBlobCache[$cacheID];
+               }
+               
+               wfDebug( "ExternalStoreDB::fetchBlob cache miss on $cacheID\n" );
+               
+               $dbr =& $this->getSlave( $cluster );
+               $ret = $dbr->selectField( 'blobs', 'blob_text', array( 'blob_id' => $id ) );
+               if( $itemID !== false ) {
+                       // Unserialise object; caller extracts item
+                       $ret = unserialize( $ret );
                }
+               
+               $wgExternalBlobCache = array( $cacheID => &$ret );
                return $ret;
        }