From: Aaron Schulz <aschulz@wikimedia.org>
Date: Thu, 12 Jul 2012 17:14:08 +0000 (-0700)
Subject: [FileBackend] Tweaks to speed up backend copy script.
X-Git-Tag: 1.31.0-rc.0~23065^2
X-Git-Url: http://git.cyclocoop.org/%22.%24redirect_annul.%22?a=commitdiff_plain;h=bd006888c9337c916ee2401600d9d8a442a1dc3f;p=lhc%2Fweb%2Fwiklou.git

[FileBackend] Tweaks to speed up backend copy script.

* Added 'missingonly' option to copy based only on a listing comparison.
* Made copy operation batches use doQuickOperations().

Change-Id: If5657a4c4d895980e27776357bb408b350727382
---

diff --git a/maintenance/copyFileBackend.php b/maintenance/copyFileBackend.php
index 8db0a7f4bd..184bf3d8ed 100644
--- a/maintenance/copyFileBackend.php
+++ b/maintenance/copyFileBackend.php
@@ -44,6 +44,7 @@ class CopyFileBackend extends Maintenance {
 		$this->addOption( 'subdir', 'Only do items in this child directory', false, true );
 		$this->addOption( 'ratefile', 'File to check periodically for batch size', false, true );
 		$this->addOption( 'skiphash', 'Skip SHA-1 sync checks for files' );
+		$this->addOption( 'missingonly', 'Only copy files missing from destination listing' );
 		$this->setBatchSize( 50 );
 	}
 
@@ -65,12 +66,32 @@ class CopyFileBackend extends Maintenance {
 				$this->output( "Doing container '$container'...\n" );
 			}
 
-			$dir = $src->getRootStoragePath() . "/$backendRel";
-			$srcPathsRel = $src->getFileList( array( 'dir' => $dir ) );
+			$srcPathsRel = $src->getFileList( array(
+				'dir' => $src->getRootStoragePath() . "/$backendRel" ) );
 			if ( $srcPathsRel === null ) {
 				$this->error( "Could not list files in $container.", 1 ); // die
 			}
 
+			// Do a listing comparison if specified
+			if ( $this->hasOption( 'missingonly' ) ) {
+				$relFilesSrc = array();
+				$relFilesDst = array();
+				foreach ( $srcPathsRel as $srcPathRel ) {
+					$relFilesSrc[] = $srcPathRel;
+				}
+				$dstPathsRel = $dst->getFileList( array(
+					'dir' => $dst->getRootStoragePath() . "/$backendRel" ) );
+				if ( $dstPathsRel === null ) {
+					$this->error( "Could not list files in $container.", 1 ); // die
+				}
+				foreach ( $dstPathsRel as $dstPathRel ) {
+					$relFilesDst[] = $dstPathRel;
+				}
+				// Only copy the missing files over in the next loop
+				$srcPathsRel = array_diff( $relFilesSrc, $relFilesDst );
+				$this->output( count( $srcPathsRel ) . " file(s) need to be copied.\n" );
+			}
+
 			$batchPaths = array();
 			foreach ( $srcPathsRel as $srcPathRel ) {
 				// Check up on the rate file periodically to adjust the concurrency
@@ -131,7 +152,7 @@ class CopyFileBackend extends Maintenance {
 		}
 
 		$t_start = microtime( true );
-		$status = $dst->doOperations( $ops, array( 'nonJournaled' => 1 ) );
+		$status = $dst->doQuickOperations( $ops );
 		$ellapsed_ms = floor( ( microtime( true ) - $t_start ) * 1000 );
 		if ( !$status->isOK() ) {
 			$this->error( print_r( $status->getErrorsArray(), true ) );