154 $endId =
$dbr->selectField(
'revision',
'MAX(rev_id)',
false, __METHOD__ );
158 echo
"Finding revisions...\n";
161 $res =
$dbr->select( [
'revision',
'text' ],
162 [
'rev_id',
'rev_page',
'old_id',
'old_flags',
'old_text' ],
164 'rev_id > ' .
$dbr->addQuotes( $startId ),
165 'rev_text_id=old_id',
167 'old_flags ' .
$dbr->buildLike(
$dbr->anyString(),
'external',
$dbr->anyString() ),
171 'ORDER BY' =>
'rev_id',
172 'LIMIT' => $this->batchSize
175 if ( !
$res->numRows() ) {
180 foreach (
$res as $row ) {
181 $startId = $row->rev_id;
184 echo
"Invalid DB:// URL in rev_id {$row->rev_id}\n";
187 if ( !in_array( $info[
'cluster'], $this->clusters ) ) {
188 echo
"Invalid cluster returned in SQL query: {$info['cluster']}\n";
192 'bt_page' => $row->rev_page,
193 'bt_rev_id' => $row->rev_id,
194 'bt_text_id' => $row->old_id,
195 'bt_cluster' => $info[
'cluster'],
196 'bt_blob_id' => $info[
'id'],
197 'bt_cgz_hash' => $info[
'hash']
199 if ( $this->doBlobOrphans ) {
200 gmp_setbit( $this->trackedBlobs[$info[
'cluster']], $info[
'id'] );
203 $dbw->insert(
'blob_tracking', $insertBatch, __METHOD__ );
204 $rowsInserted += count( $insertBatch );
207 if ( $batchesDone >= $this->reportingInterval ) {
209 echo
"$startId / $endId\n";
213 echo
"Found $rowsInserted revisions\n";
222 # Wait until the blob_tracking table is available in the replica DB
225 $pos = $dbw->getMasterPos();
226 $dbr->masterPosWait( $pos, 100000 );
230 $endId =
$dbr->selectField(
'text',
'MAX(old_id)',
false, __METHOD__ );
234 echo
"Finding orphan text...\n";
236 # Scan the text table for orphan text
238 $res =
$dbr->select( [
'text',
'blob_tracking' ],
239 [
'old_id',
'old_flags',
'old_text' ],
241 'old_id>' .
$dbr->addQuotes( $startId ),
243 'old_flags ' .
$dbr->buildLike(
$dbr->anyString(),
'external',
$dbr->anyString() ),
248 'ORDER BY' =>
'old_id',
249 'LIMIT' => $this->batchSize
251 [
'blob_tracking' => [
'LEFT JOIN',
'bt_text_id=old_id' ] ]
254 foreach (
$res as $row ) {
255 $ids[] = $row->old_id;
258 if ( !
$res->numRows() ) {
263 foreach (
$res as $row ) {
264 $startId = $row->old_id;
267 echo
"Invalid DB:// URL in old_id {$row->old_id}\n";
270 if ( !in_array( $info[
'cluster'], $this->clusters ) ) {
271 echo
"Invalid cluster returned in SQL query\n";
278 'bt_text_id' => $row->old_id,
279 'bt_cluster' => $info[
'cluster'],
280 'bt_blob_id' => $info[
'id'],
281 'bt_cgz_hash' => $info[
'hash']
283 if ( $this->doBlobOrphans ) {
284 gmp_setbit( $this->trackedBlobs[$info[
'cluster']], $info[
'id'] );
287 $dbw->insert(
'blob_tracking', $insertBatch, __METHOD__ );
289 $rowsInserted += count( $insertBatch );
291 if ( $batchesDone >= $this->reportingInterval ) {
293 echo
"$startId / $endId\n";
297 echo
"Found $rowsInserted orphan text rows\n";
308 if ( !extension_loaded(
'gmp' ) ) {
309 echo
"Can't find orphan blobs, need bitfield support provided by GMP.\n";
316 foreach ( $this->clusters
as $cluster ) {
317 echo
"Searching for orphan blobs in $cluster...\n";
322 if ( strpos(
$e->error,
'Unknown database' ) !==
false ) {
323 echo
"No database on $cluster\n";
325 echo
"Error on $cluster: " .
$e->getMessage() .
"\n";
329 $table = $extDB->getLBInfo(
'blobs table' );
330 if ( is_null( $table ) ) {
333 if ( !$extDB->tableExists( $table ) ) {
334 echo
"No blobs table on cluster $cluster\n";
339 $actualBlobs = gmp_init( 0 );
340 $endId = $extDB->selectField( $table,
'MAX(blob_id)',
false, __METHOD__ );
344 $res = $extDB->select( $table,
346 [
'blob_id > ' . $extDB->addQuotes( $startId ) ],
348 [
'LIMIT' => $this->batchSize,
'ORDER BY' =>
'blob_id' ]
351 if ( !
$res->numRows() ) {
355 foreach (
$res as $row ) {
356 gmp_setbit( $actualBlobs, $row->blob_id );
358 $startId = $row->blob_id;
361 if ( $batchesDone >= $this->reportingInterval ) {
363 echo
"$startId / $endId\n";
369 $orphans = gmp_and( $actualBlobs, gmp_com( $this->trackedBlobs[$cluster] ) );
376 $id = gmp_scan1( $orphans, $id );
381 'bo_cluster' => $cluster,
384 if ( count( $insertBatch ) > $this->batchSize ) {
385 $dbw->insert(
'blob_orphans', $insertBatch, __METHOD__ );
392 if ( $insertBatch ) {
393 $dbw->insert(
'blob_orphans', $insertBatch, __METHOD__ );
395 echo
"Found $numOrphans orphan(s) in $cluster\n";