From: Tim Starling <tstarling@users.mediawiki.org>
Date: Fri, 19 Dec 2008 08:43:26 +0000 (+0000)
Subject: Stats on the blob_orphans table, to make sure we're not losing anything by dropping... 
X-Git-Tag: 1.31.0-rc.0~43882
X-Git-Url: https://git.heureux-cyclage.org/w/index.php?a=commitdiff_plain;h=be768f01fd00f7e0c67997ae00aca116f2bf3ef2;p=lhc%2Fweb%2Fwiklou.git

Stats on the blob_orphans table, to make sure we're not losing anything by dropping the tracked orphans.
---

diff --git a/maintenance/storage/orphanStats.php b/maintenance/storage/orphanStats.php
new file mode 100644
index 0000000000..77bf465d08
--- /dev/null
+++ b/maintenance/storage/orphanStats.php
@@ -0,0 +1,44 @@
+<?php
+
+/**
+ * Show some statistics on the blob_orphans table, created with trackBlobs.php
+ */
+require_once( dirname(__FILE__).'/../commandLine.inc' );
+
+$stats = new OrphanStats;
+$stats->execute();
+
+class OrphanStats {
+	function getDB( $cluster ) {
+		$lb = wfGetLBFactory()->getExternalLB( $cluster );
+		return $lb->getConnection( DB_SLAVE );
+	}
+
+	function execute() {
+		$extDBs = array();
+		$dbr = wfGetDB( DB_SLAVE );
+		$res = $dbr->select( 'blob_orphans', '*', false, __METHOD__ );
+		
+		$num = 0;
+		$totalSize = 0;
+		$hashes = array();
+		$maxSize = 0;
+
+		foreach ( $res as $boRow ) {
+			$extDB = $this->getDB( $boRow->bo_cluster );
+			$blobRow = $extDB->selectRow( 'blobs', '*', array( 'blob_id' => $boRow->bo_blob_id ), __METHOD__ );
+			
+			$num++;
+			$size = strlen( $blobRow->blob_text );
+			$totalSize += $size;
+			$hashes[ sha1( $blobRow->blob_text ) ] = true;
+			$maxSize = max( $size, $maxSize );
+		}
+		unset( $res );
+
+		echo "Number of orphans: $num\n" .
+			"Average size: " . round( $totalSize / $num, 0 ) . " bytes\n" .
+			"Max size: $maxSize\n" . 
+			"Number of unique texts: " . count( $hashes ) . "\n";
+	}
+}