aboutsummaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
Diffstat
-rw-r--r--maintenance/storage/blobs.sql8
-rw-r--r--maintenance/storage/compressOld.inc246
-rw-r--r--maintenance/storage/compressOld.php81
-rw-r--r--maintenance/storage/dumpRev.php14
-rw-r--r--maintenance/storage/moveToExternal.php80
-rw-r--r--maintenance/storage/resolveStubs.php108
6 files changed, 537 insertions, 0 deletions
diff --git a/maintenance/storage/blobs.sql b/maintenance/storage/blobs.sql
new file mode 100644
index 000000000000..3e72f77f3339
--- /dev/null
+++ b/maintenance/storage/blobs.sql
@@ -0,0 +1,8 @@
+-- Blobs table for external storage
+
+CREATE TABLE /*$wgDBprefix*/blobs (
+ blob_id int(8) NOT NULL default '0' AUTO_INCREMENT,
+ blob_text mediumtext,
+ PRIMARY KEY (blob_id)
+) TYPE=InnoDB;
+
diff --git a/maintenance/storage/compressOld.inc b/maintenance/storage/compressOld.inc
new file mode 100644
index 000000000000..344a490f2875
--- /dev/null
+++ b/maintenance/storage/compressOld.inc
@@ -0,0 +1,246 @@
+<?php
+/**
+ * @package MediaWiki
+ * @subpackage Maintenance
+ */
+
+/** */
+require_once( 'Revision.php' );
+require_once( 'ExternalStoreDB.php' );
+
+/** @todo document */
+function compressOldPages( $start = 0, $extdb = '' ) {
+ $fname = 'compressOldPages';
+
+ $chunksize = 50;
+ print "Starting from old_id $start...\n";
+ $dbw =& wfGetDB( DB_MASTER );
+ do {
+ $end = $start + $chunksize;
+ $res = $dbw->select( 'text', array( 'old_id','old_flags','old_namespace','old_title','old_text' ),
+ "old_id>=$start", $fname, array( 'ORDER BY' => 'old_id', 'LIMIT' => $chunksize, 'FOR UPDATE' ) );
+ if( $dbw->numRows( $res ) == 0 ) {
+ break;
+ }
+ $last = $start;
+ while( $row = $dbw->fetchObject( $res ) ) {
+ # print " {$row->old_id} - {$row->old_namespace}:{$row->old_title}\n";
+ compressPage( $row, $extdb );
+ $last = $row->old_id;
+ }
+ $dbw->freeResult( $res );
+ $start = $last + 1; # Deletion may leave long empty stretches
+ print "$start...\n";
+ } while( true );
+}
+
+/** @todo document */
+function compressPage( $row, $extdb ) {
+ $fname = 'compressPage';
+ if ( false !== strpos( $row->old_flags, 'gzip' ) || false !== strpos( $row->old_flags, 'object' ) ) {
+ #print "Already compressed row {$row->old_id}\n";
+ return false;
+ }
+ $dbw =& wfGetDB( DB_MASTER );
+ $flags = $row->old_flags ? "{$row->old_flags},gzip" : "gzip";
+ $compress = gzdeflate( $row->old_text );
+
+ # Store in external storage if required
+ if ( $extdb !== '' ) {
+ $storeObj = new ExternalStoreDB;
+ $compress = $storeObj->store( $extdb, $compress );
+ if ( $compress === false ) {
+ print "Unable to store object\n";
+ return false;
+ }
+ }
+
+ # Update text row
+ $dbw->update( 'text',
+ array( /* SET */
+ 'old_flags' => $flags,
+ 'old_text' => $compress
+ ), array( /* WHERE */
+ 'old_id' => $row->old_id
+ ), $fname, 'LIMIT 1'
+ );
+ return true;
+}
+
+define( 'LS_INDIVIDUAL', 0 );
+define( 'LS_CHUNKED', 1 );
+
+/** @todo document */
+function compressWithConcat( $startId, $maxChunkSize, $maxChunkFactor, $factorThreshold, $beginDate, $endDate )
+{
+ $fname = 'compressWithConcat';
+ $loadStyle = LS_CHUNKED;
+
+ $dbr =& wfGetDB( DB_SLAVE );
+ $dbw =& wfGetDB( DB_MASTER );
+
+ # Get all articles by page_id
+ $maxPageId = $dbr->selectField( 'page', 'max(page_id)', '', $fname );
+ $pageConds = array();
+
+ /*
+ if ( $exclude_ns0 ) {
+ print "Excluding main namespace\n";
+ $pageConds[] = 'page_namespace<>0';
+ }
+ if ( $queryExtra ) {
+ $pageConds[] = $queryExtra;
+ }
+ */
+
+ # For each article, get a list of revisions which fit the criteria
+ # No recompression, use a condition on old_flags
+ $conds = array(
+ "old_flags NOT LIKE '%object%' " .
+ " AND (old_flags NOT LIKE '%external%' OR old_text NOT LIKE 'DB://%/%/%')");
+
+ if ( $beginDate ) {
+ $conds[] = "rev_timestamp>'" . $beginDate . "'";
+ }
+ if ( $endDate ) {
+ $conds[] = "rev_timestamp<'" . $endDate . "'";
+ }
+ if ( $loadStyle == LS_CHUNKED ) {
+ $tables = array( 'revision', 'text' );
+ $fields = array( 'rev_id', 'rev_text_id', 'old_flags', 'old_text' );
+ $conds[] = 'rev_text_id=old_id';
+ $revLoadOptions = 'FOR UPDATE';
+ } else {
+ $tables = array( 'revision' );
+ $fields = array( 'rev_id', 'rev_text_id' );
+ $revLoadOptions = array();
+ }
+
+ $oldReadsSinceLastSlaveWait = 0; #check slave lag periodically
+ $totalMatchingRevisions = 0;
+ $masterPos = false;
+ for ( $pageId = $startId; $pageId <= $maxPageId; $pageId++ ) {
+ $pageRes = $dbr->select( 'page', array('page_id', 'page_namespace', 'page_title'),
+ $pageConds + array('page_id' => $pageId), $fname );
+ if ( $dbr->numRows( $pageRes ) == 0 ) {
+ continue;
+ }
+ $pageRow = $dbr->fetchObject( $pageRes );
+
+ # Display progress
+ $titleObj = Title::makeTitle( $pageRow->page_namespace, $pageRow->page_title );
+ print "$pageId\t" . $titleObj->getPrefixedDBkey() . " ";
+
+ # Load revisions
+ $revRes = $dbw->select( $tables, $fields,
+ array( 'rev_page' => $pageRow->page_id ) + $conds,
+ $fname,
+ $revLoadOptions
+ );
+ $revs = array();
+ while ( $revRow = $dbw->fetchObject( $revRes ) ) {
+ $revs[] = $revRow;
+ }
+
+ if ( count( $revs ) < 2) {
+ # No revisions matching, no further processing
+ print "\n";
+ continue;
+ }
+
+ # For each chunk
+ $i = 0;
+ while ( $i < count( $revs ) ) {
+ if ( $i < count( $revs ) - $maxChunkSize ) {
+ $thisChunkSize = $maxChunkSize;
+ } else {
+ $thisChunkSize = count( $revs ) - $i;
+ }
+
+ $chunk = new ConcatenatedGzipHistoryBlob();
+ $stubs = array();
+ $dbw->begin();
+ $usedChunk = false;
+ $primaryOldid = $revs[$i]->rev_text_id;
+
+ # Get the text of each revision and add it to the object
+ for ( $j = 0; $j < $thisChunkSize && $chunk->isHappy( $maxChunkFactor, $factorThreshold ); $j++ ) {
+ $oldid = $revs[$i + $j]->rev_text_id;
+
+ # Get text
+ if ( $loadStyle == LS_INDIVIDUAL ) {
+ $textRow = $dbw->selectRow( 'text',
+ array( 'old_flags', 'old_text' ),
+ array( 'old_id' => $oldid ),
+ $fname,
+ 'FOR UPDATE'
+ );
+ $text = Revision::getRevisionText( $textRow );
+ } else {
+ $text = Revision::getRevisionText( $revs[$i + $j] );
+ }
+
+ if ( $text === false ) {
+ print "\nError, unable to get text in old_id $oldid\n";
+ #$dbw->delete( 'old', array( 'old_id' => $oldid ) );
+ }
+
+ if ( $j == 0 ) {
+ $chunk->setText( $text );
+ print '.';
+ } else {
+ # Don't make a stub if it's going to be longer than the article
+ # Stubs are typically about 100 bytes
+ if ( strlen( $text ) < 120 ) {
+ $stub = false;
+ print 'x';
+ } else {
+ $stub = $chunk->addItem( $text );
+ $stub->setLocation( $primaryOldid );
+ $hash = $stub->getHash();
+ $stub = serialize( $stub );
+ print '.';
+ $usedChunk = true;
+ }
+ $stubs[$j] = $stub;
+ }
+ }
+ $thisChunkSize = $j;
+
+ # If we couldn't actually use any stubs because the pages were too small, do nothing
+ if ( $usedChunk ) {
+ # Store the main object
+ $dbw->update( 'text',
+ array( /* SET */
+ 'old_text' => serialize( $chunk ),
+ 'old_flags' => 'object,utf-8',
+ ), array( /* WHERE */
+ 'old_id' => $primaryOldid
+ )
+ );
+
+ # Store the stub objects
+ for ( $j = 1; $j < $thisChunkSize; $j++ ) {
+ # Skip if not compressing
+ if ( $stubs[$j] !== false ) {
+ $dbw->update( 'text',
+ array( /* SET */
+ 'old_text' => $stubs[$j],
+ 'old_flags' => 'object,utf-8',
+ ), array( /* WHERE */
+ 'old_id' => $revs[$i + $j]->rev_text_id
+ )
+ );
+ }
+ }
+ }
+ # Done, next
+ print "/";
+ $dbw->commit();
+ $i += $thisChunkSize;
+ }
+ print "\n";
+ }
+ return true;
+}
+?>
diff --git a/maintenance/storage/compressOld.php b/maintenance/storage/compressOld.php
new file mode 100644
index 000000000000..35d8c829460e
--- /dev/null
+++ b/maintenance/storage/compressOld.php
@@ -0,0 +1,81 @@
+<?php
+/**
+ * Compress the text of a wiki
+ *
+ * @package MediaWiki
+ * @subpackage Maintenance
+ */
+
+/** */
+
+/**
+ * Usage:
+ *
+ * Non-wikimedia
+ * php compressOld.php [options...]
+ *
+ * Wikimedia
+ * php compressOld.php <database> [options...]
+ *
+ * Options are:
+ * -t <type> set compression type to either:
+ * gzip: compress revisions independently
+ * concat: concatenate revisions and compress in chunks (default)
+ * -c <chunk-size> maximum number of revisions in a concat chunk
+ * -b <begin-date> earliest date to check for uncompressed revisions
+ * -e <end-date> latest revision date to compress
+ * -s <start-id> the old_id to start from
+ * -f <max-factor> the maximum ratio of compressed chunk bytes to uncompressed avg. revision bytes
+ * -h <threshold> is a minimum number of KB, where <max-factor> cuts in
+ * --extdb <cluster> store specified revisions in an external cluster (untested)
+ *
+ */
+
+$optionsWithArgs = array( 't', 'c', 's', 'f', 'h', 'extdb' );
+require_once( "../commandLine.inc" );
+require_once( "compressOld.inc" );
+
+if( !function_exists( "gzdeflate" ) ) {
+ print "You must enable zlib support in PHP to compress old revisions!\n";
+ print "Please see http://www.php.net/manual/en/ref.zlib.php\n\n";
+ die();
+}
+
+$defaults = array(
+ 't' => 'concat',
+ 'c' => 20,
+ 's' => 0,
+ 'f' => 3,
+ 'h' => 100,
+ 'b' => '',
+ 'e' => '',
+ 'extdb' => '',
+);
+
+$options = $options + $defaults;
+
+if ( $options['t'] != 'concat' && $options['t'] != 'gzip' ) {
+ print "Type \"{$options['t']}\" not supported\n";
+}
+
+print "Depending on the size of your database this may take a while!\n";
+print "If you abort the script while it's running it shouldn't harm anything,\n";
+print "but if you haven't backed up your data, you SHOULD abort now!\n\n";
+print "Press control-c to abort first (will proceed automatically in 5 seconds)\n";
+#sleep(5);
+
+$success = true;
+if ( $options['t'] == 'concat' ) {
+ $success = compressWithConcat( $options['s'], $options['c'], $options['f'], $options['h'], $options['b'],
+ $options['e'], $options['extdb'] );
+} else {
+ compressOldPages( $options['s'], $options['extdb'] );
+}
+
+if ( $success ) {
+ print "Done.\n";
+}
+
+exit();
+
+?>
diff --git a/maintenance/storage/dumpRev.php b/maintenance/storage/dumpRev.php
new file mode 100644
index 000000000000..4d0ccb5853af
--- /dev/null
+++ b/maintenance/storage/dumpRev.php
@@ -0,0 +1,14 @@
+<?php
+
+require_once( 'commandLine.inc' );
+$dbr =& wfGetDB( DB_SLAVE );
+$row = $dbr->selectRow( 'old', array( 'old_flags', 'old_text' ), array( 'old_id' => $args[0] ) );
+$obj = unserialize( $row->old_text );
+
+if ( get_class( $obj ) == 'concatenatedgziphistoryblob' ) {
+ print_r( array_keys( $obj->mItems ) );
+} else {
+ var_dump( $obj );
+}
+
+?>
diff --git a/maintenance/storage/moveToExternal.php b/maintenance/storage/moveToExternal.php
new file mode 100644
index 000000000000..8b441d2023bd
--- /dev/null
+++ b/maintenance/storage/moveToExternal.php
@@ -0,0 +1,80 @@
+<?php
+
+define( 'REPORTING_INTERVAL', 100 );
+define( 'STUB_HEADER', 'O:15:"historyblobstub"' );
+
+if ( !defined( 'MEDIAWIKI' ) ) {
+ $optionsWithArgs = array( 'm' );
+
+ require_once( '../commandLine.inc' );
+ require_once( 'ExternalStoreDB.php' );
+ require_once( 'resolveStubs.php' );
+
+ $fname = 'moveToExternal';
+
+ if ( !isset( $args[0] ) ) {
+ print "Usage: php moveToExternal.php [-m <maxid>] <cluster>\n";
+ exit;
+ }
+
+ $cluster = $args[0];
+ $dbw =& wfGetDB( DB_MASTER );
+
+ if ( isset( $options['m'] ) ) {
+ $maxID = $options['m'];
+ } else {
+ $maxID = $dbw->selectField( 'text', 'MAX(old_id)', false, $fname );
+ }
+
+ moveToExternal( $cluster, $maxID );
+}
+
+
+
+function moveToExternal( $cluster, $maxID ) {
+ $fname = 'moveToExternal';
+ $dbw =& wfGetDB( DB_MASTER );
+
+ print "Moving $maxID text rows to external storage\n";
+ $ext = new ExternalStoreDB;
+ for ( $id = 1; $id <= $maxID; $id++ ) {
+ if ( !($id % REPORTING_INTERVAL) ) {
+ print "$id\n";
+ wfWaitForSlaves( 5 );
+ }
+ $row = $dbw->selectRow( 'text', array( 'old_flags', 'old_text' ),
+ array(
+ 'old_id' => $id,
+ "old_flags NOT LIKE '%external%'",
+ ), $fname );
+ if ( !$row ) {
+ # Non-existent or already done
+ continue;
+ }
+
+ # Resolve stubs
+ $flags = explode( ',', $row->old_flags );
+ if ( in_array( 'object', $flags )
+ && substr( $row->old_text, 0, strlen( STUB_HEADER ) ) === STUB_HEADER )
+ {
+ resolveStub( $id, $row->old_text, $row->old_flags );
+ continue;
+ }
+
+ $url = $ext->store( $cluster, $row->old_text );
+ if ( !$url ) {
+ print "Error writing to external storage\n";
+ exit;
+ }
+ if ( $row->old_flags === '' ) {
+ $flags = 'external';
+ } else {
+ $flags = "{$row->old_flags},external";
+ }
+ $dbw->update( 'text',
+ array( 'old_flags' => $flags, 'old_text' => $url ),
+ array( 'old_id' => $id ), $fname );
+ }
+}
+
+?>
diff --git a/maintenance/storage/resolveStubs.php b/maintenance/storage/resolveStubs.php
new file mode 100644
index 000000000000..d22976c5f75d
--- /dev/null
+++ b/maintenance/storage/resolveStubs.php
@@ -0,0 +1,108 @@
+<?php
+
+define( 'REPORTING_INTERVAL', 100 );
+
+if ( !defined( 'MEDIAWIKI' ) ) {
+ $optionsWithArgs = array( 'm' );
+
+ require_once( '../commandLine.inc' );
+ require_once( 'includes/ExternalStoreDB.php' );
+
+ resolveStubs();
+}
+
+/**
+ * Convert history stubs that point to an external row to direct
+ * external pointers
+ */
+function resolveStubs() {
+ $fname = 'resolveStubs';
+
+ print "Retrieving stub rows...\n";
+ $dbr =& wfGetDB( DB_SLAVE );
+ $maxID = $dbr->selectField( 'text', 'MAX(old_id)', false, $fname );
+ $stubs = array();
+ $flagsArray = array();
+
+ # Do it in 100 blocks
+ for ( $b = 0; $b < 100; $b++ ) {
+ print "$b%\r";
+ $start = intval($maxID / 100) * $b + 1;
+ $end = intval($maxID / 100) * ($b + 1);
+
+ $res = $dbr->select( 'text', array( 'old_id', 'old_text', 'old_flags' ),
+ "old_id>=$start AND old_id<=$end AND old_flags like '%object%' ".
+ "AND old_text LIKE 'O:15:\"historyblobstub\"%'", $fname );
+ while ( $row = $dbr->fetchObject( $res ) ) {
+ $stubs[$row->old_id] = $row->old_text;
+ $flagsArray[$row->old_id] = $row->old_flags;
+ }
+ $dbr->freeResult( $res );
+ }
+ print "100%\n";
+
+ print "\nConverting " . count( $stubs ) . " rows ...\n";
+
+ # Get master database, no transactions
+ $dbw =& wfGetDB( DB_MASTER );
+ $dbw->clearFlag( DBO_TRX );
+ $dbw->immediateCommit();
+
+ $i = 0;
+ foreach( $stubs as $id => $stub ) {
+ if ( !(++$i % REPORTING_INTERVAL) ) {
+ print "$i\n";
+ wfWaitForSlaves( 5 );
+ }
+
+ resolveStub( $id, $stub, $flagsArray[$id] );
+ }
+}
+
+/**
+ * Resolve a history stub
+ */
+function resolveStub( $id, $stubText, $flags ) {
+ $fname = 'resolveStub';
+
+ $stub = unserialize( $stubText );
+ $flags = explode( ',', $flags );
+
+ $dbr =& wfGetDB( DB_SLAVE );
+ $dbw =& wfGetDB( DB_MASTER );
+
+ if ( get_class( $stub ) !== 'historyblobstub' ) {
+ print "Error, invalid stub object\n";
+ return;
+ }
+
+ # Get the (maybe) external row
+ $externalRow = $dbr->selectRow( 'text', array( 'old_text' ),
+ array( 'old_id' => $stub->mOldId, "old_flags LIKE '%external%'" ),
+ $fname
+ );
+
+ if ( !$externalRow ) {
+ # Object wasn't external
+ continue;
+ }
+
+ # Preserve the legacy encoding flag, but switch from object to external
+ if ( in_array( 'utf-8', $flags ) ) {
+ $newFlags = 'external,utf-8';
+ } else {
+ $newFlags = 'external';
+ }
+
+ # Update the row
+ $dbw->update( 'text',
+ array( /* SET */
+ 'old_flags' => $newFlags,
+ 'old_text' => $externalRow->old_text . '/' . $stub->mHash
+ ),
+ array( /* WHERE */
+ 'old_id' => $id
+ ), $fname
+ );
+}
+?>
generated by cgit v1.2.3 (git 2.25.1) at 2025-04-21 14:13:14 +0000