- Patch #19298 by Jeremy: loose caching!

Drupal's existing caching mechanism doesn't perform well on highly dynamic websites in which the cache is flushed frequently. One example is a site that is under attack by a spambot that is posting spam comments every few seconds, causing all cached pages to be flushed every few seconds.  Loose caching immediately flushes the cache only for specific users who have modified cached data (whether or not they are logged in), delaying the flushing of data for other users by several minutes.

(I rewrote the help text a bit and made minor changes to the code comments.)

2 files changed, 62 insertions, 6 deletions

diff --git a/includes/bootstrap.inc b/includes/bootstrap.inc
index 7ad4626a3..def6e412c 100644
--- a/includes/bootstrap.inc
+++ b/includes/bootstrap.inc
@@ -9,6 +9,10 @@
 define('CACHE_PERMANENT', 0);
 define('CACHE_TEMPORARY', -1);
 
+define('CACHE_DISABLED', 0);
+define('CACHE_ENABLED_STRICT', 1);
+define('CACHE_ENABLED_LOOSE', 2);
+
 define('WATCHDOG_NOTICE', 0);
 define('WATCHDOG_WARNING', 1);
 define('WATCHDOG_ERROR', 2);
@@ -196,9 +200,36 @@ function variable_del($name) {
  *   The cache ID of the data to retrieve.
  */
 function cache_get($key) {
-  $cache = db_fetch_object(db_query("SELECT data, created, headers FROM {cache} WHERE cid = '%s'", $key));
+  global $user;
+  $sid = session_id();
+
+  // CACHE_ENABLED_LOOSE garbage collection
+  $cache_flush = variable_get('cache_flush', 0);
+  if ($cache_flush && ($cache_flush + variable_get('cache_flush_delay', 300) <= time())) {
+    // Time to flush old cache data
+    db_query("DELETE FROM {cache} WHERE expire != %d AND expire <= %d", CACHE_PERMANENT, $cache_flush);
+    variable_set('cache_flush', 0);
+  }
+
+  $cache = db_fetch_object(db_query("SELECT data, created, headers, expire FROM {cache} WHERE cid = '%s'", $key));
   if (isset($cache->data)) {
-    $cache->data = db_decode_blob($cache->data);
+    // If data is permanent or using strict caching, always return data.
+    if ($cache->expire == CACHE_PERMANENT || variable_get('cache', CACHE_DISABLED) == CACHE_ENABLED_STRICT) {
+      $cache->data = db_decode_blob($cache->data);
+    }
+    // If using loose caching, validate data is current before we return it by
+    // making sure the cache entry was created before the timestamp in the
+    // current session's cache timer.  The cache variable is already loaded
+    // into the $user object by sess_read in session.inc.
+    else {
+      if ($user->cache > $cache->created) {
+        // This cache data is too old and thus not valid for us, ignore it.
+        return 0;
+      }
+      else {
+        $cache->data = db_decode_blob($cache->data);
+      }
+    }
     return $cache;
   }
   return 0;
@@ -235,16 +266,41 @@ function cache_set($cid, $data, $expire = CACHE_PERMANENT, $headers = NULL) {
  * Expire data from the cache.
  *
  * @param $cid
- *   If set, the cache ID to delete. Otherwise, all cache entries that can expire
- *   are deleted.
+ *   If set, the cache ID to delete. Otherwise, all cache entries that can
+ *   expire are deleted.
  *
  * @param $wildcard
  *   If set to true, the $cid is treated as a substring to match rather than a
  *   complete ID.
  */
 function cache_clear_all($cid = NULL, $wildcard = false) {
+  global $user;
+  $sid = session_id();
+
   if (empty($cid)) {
-    db_query("DELETE FROM {cache} WHERE expire != %d AND expire < %d", CACHE_PERMANENT, time());
+    if (variable_get('cache', CACHE_DISABLED) == CACHE_ENABLED_STRICT) {
+      // Strict caching, flush all temporary cache entries:
+      db_query("DELETE FROM {cache} WHERE expire != %d AND expire < %d", CACHE_PERMANENT, time());
+    }
+    else {
+      $cache_flush = variable_get('cache_flush', 0);
+      // Loose caching, only flush temporary cache entries that have been
+      // invalidated for more than maximum allowable time.
+      if ($cache_flush && ($cache_flush + variable_get('cache_flush_delay', 300) <= time())) {
+        // Only flush cache data older than $cache_flush, as newer data may
+        // now be valid.
+        db_query("DELETE FROM {cache} WHERE expire != %d AND expire <= %d", CACHE_PERMANENT, $cache_flush);
+        variable_set('cache_flush', 0);
+      }
+      // Invalidate temporary cache data only for current user/session.  We
+      // set $user->cache, which gets saved into the sessions table by
+      // sess_write() in session.inc.
+      $user->cache = time();
+      if (variable_get('cache_flush', 0) == 0) {
+        // Set timestamp to know which cache entries we eventually clear:
+        variable_set('cache_flush', time());
+      }
+    }
   }
   else {
     if ($wildcard) {
diff --git a/includes/session.inc b/includes/session.inc
index 73514e37d..517651704 100644
--- a/includes/session.inc
+++ b/includes/session.inc
@@ -45,7 +45,7 @@ function sess_read($key) {
 function sess_write($key, $value) {
   global $user;
 
-  db_query("UPDATE {sessions} SET uid = %d, hostname = '%s', session = '%s', timestamp = %d WHERE sid = '%s'", $user->uid, $_SERVER["REMOTE_ADDR"], $value, time(), $key);
+  db_query("UPDATE {sessions} SET uid = %d, cache = %d, hostname = '%s', session = '%s', timestamp = %d WHERE sid = '%s'", $user->uid, $user->cache, $_SERVER["REMOTE_ADDR"], $value, time(), $key);
 
   return '';
 }