orphanStats.php

Go to the documentation of this file.

<?php
require_once __DIR__ . '/../Maintenance.php';
 
use MediaWiki\MediaWikiServices;
 
class OrphanStats extends Maintenance {
    public function __construct() {
        parent::__construct();
        $this->addDescription(
            "Show some statistics on the blob_orphans table, created with trackBlobs.php" );
    }
 
    protected function getDB( $cluster, $groups = [], $wiki = false ) {
        $lbFactory = MediaWikiServices::getInstance()->getDBLoadBalancerFactory();
        $lb = $lbFactory->getExternalLB( $cluster );
 
        return $lb->getMaintenanceConnectionRef( DB_REPLICA );
    }
 
    public function execute() {
        $dbr = $this->getDB( DB_REPLICA );
        if ( !$dbr->tableExists( 'blob_orphans' ) ) {
            $this->fatalError( "blob_orphans doesn't seem to exist, need to run trackBlobs.php first" );
        }
        $res = $dbr->select( 'blob_orphans', '*', '', __METHOD__ );
 
        $num = 0;
        $totalSize = 0;
        $hashes = [];
        $maxSize = 0;
 
        foreach ( $res as $row ) {
            $extDB = $this->getDB( $row->bo_cluster );
            $blobRow = $extDB->selectRow(
                'blobs',
                '*',
                [ 'blob_id' => $row->bo_blob_id ],
                __METHOD__
            );
 
            $num++;
            $size = strlen( $blobRow->blob_text );
            $totalSize += $size;
            $hashes[sha1( $blobRow->blob_text )] = true;
            $maxSize = max( $size, $maxSize );
        }
        unset( $res );
 
        $this->output( "Number of orphans: $num\n" );
        if ( $num > 0 ) {
            $this->output( "Average size: " . round( $totalSize / $num, 0 ) . " bytes\n" .
                "Max size: $maxSize\n" .
                "Number of unique texts: " . count( $hashes ) . "\n" );
        }
    }
}
 
$maintClass = OrphanStats::class;
require_once RUN_MAINTENANCE_IF_MAIN;