MediaWiki  master
syncFileBackend.php
Go to the documentation of this file.
1 <?php
24 require_once __DIR__ . '/Maintenance.php';
25 
33  public function __construct() {
34  parent::__construct();
35  $this->addDescription( 'Sync one file backend with another using the journal' );
36  $this->addOption( 'src', 'Name of backend to sync from', true, true );
37  $this->addOption( 'dst', 'Name of destination backend to sync', false, true );
38  $this->addOption( 'start', 'Starting journal ID', false, true );
39  $this->addOption( 'end', 'Ending journal ID', false, true );
40  $this->addOption( 'posdir', 'Directory to read/record journal positions', false, true );
41  $this->addOption( 'posdump', 'Just dump current journal position into the position dir.' );
42  $this->addOption( 'postime', 'For position dumps, get the ID at this time', false, true );
43  $this->addOption( 'backoff', 'Stop at entries younger than this age (sec).', false, true );
44  $this->addOption( 'verbose', 'Verbose mode', false, false, 'v' );
45  $this->setBatchSize( 50 );
46  }
47 
48  public function execute() {
49  $src = FileBackendGroup::singleton()->get( $this->getOption( 'src' ) );
50 
51  $posDir = $this->getOption( 'posdir' );
52  if ( $posDir != '' ) {
53  $posFile = "$posDir/" . rawurlencode( $src->getDomainId() );
54  } else {
55  $posFile = false;
56  }
57 
58  if ( $this->hasOption( 'posdump' ) ) {
59  // Just dump the current position into the specified position dir
60  if ( !$this->hasOption( 'posdir' ) ) {
61  $this->fatalError( "Param posdir required!" );
62  }
63  if ( $this->hasOption( 'postime' ) ) {
64  $id = (int)$src->getJournal()->getPositionAtTime( $this->getOption( 'postime' ) );
65  $this->output( "Requested journal position is $id.\n" );
66  } else {
67  $id = (int)$src->getJournal()->getCurrentPosition();
68  $this->output( "Current journal position is $id.\n" );
69  }
70  if ( file_put_contents( $posFile, $id, LOCK_EX ) !== false ) {
71  $this->output( "Saved journal position file.\n" );
72  } else {
73  $this->output( "Could not save journal position file.\n" );
74  }
75  if ( $this->isQuiet() ) {
76  print $id; // give a single machine-readable number
77  }
78 
79  return;
80  }
81 
82  if ( !$this->hasOption( 'dst' ) ) {
83  $this->fatalError( "Param dst required!" );
84  }
85  $dst = FileBackendGroup::singleton()->get( $this->getOption( 'dst' ) );
86 
87  $start = $this->getOption( 'start', 0 );
88  if ( !$start && $posFile && is_dir( $posDir ) ) {
89  $start = is_file( $posFile )
90  ? (int)trim( file_get_contents( $posFile ) )
91  : 0;
92  ++$start; // we already did this ID, start with the next one
93  $startFromPosFile = true;
94  } else {
95  $startFromPosFile = false;
96  }
97 
98  if ( $this->hasOption( 'backoff' ) ) {
99  $time = time() - $this->getOption( 'backoff', 0 );
100  $end = (int)$src->getJournal()->getPositionAtTime( $time );
101  } else {
102  $end = $this->getOption( 'end', INF );
103  }
104 
105  $this->output( "Synchronizing backend '{$dst->getName()}' to '{$src->getName()}'...\n" );
106  $this->output( "Starting journal position is $start.\n" );
107  if ( is_finite( $end ) ) {
108  $this->output( "Ending journal position is $end.\n" );
109  }
110 
111  // Periodically update the position file
112  $callback = function ( $pos ) use ( $startFromPosFile, $posFile, $start ) {
113  if ( $startFromPosFile && $pos >= $start ) { // successfully advanced
114  file_put_contents( $posFile, $pos, LOCK_EX );
115  }
116  };
117 
118  // Actually sync the dest backend with the reference backend
119  $lastOKPos = $this->syncBackends( $src, $dst, $start, $end, $callback );
120 
121  // Update the sync position file
122  if ( $startFromPosFile && $lastOKPos >= $start ) { // successfully advanced
123  if ( file_put_contents( $posFile, $lastOKPos, LOCK_EX ) !== false ) {
124  $this->output( "Updated journal position file.\n" );
125  } else {
126  $this->output( "Could not update journal position file.\n" );
127  }
128  }
129 
130  if ( $lastOKPos === false ) {
131  if ( !$start ) {
132  $this->output( "No journal entries found.\n" );
133  } else {
134  $this->output( "No new journal entries found.\n" );
135  }
136  } else {
137  $this->output( "Stopped synchronization at journal position $lastOKPos.\n" );
138  }
139 
140  if ( $this->isQuiet() ) {
141  print $lastOKPos; // give a single machine-readable number
142  }
143  }
144 
156  protected function syncBackends(
157  FileBackend $src, FileBackend $dst, $start, $end, Closure $callback
158  ) {
159  $lastOKPos = 0; // failed
160  $first = true; // first batch
161 
162  if ( $start > $end ) { // sanity
163  $this->fatalError( "Error: given starting ID greater than ending ID." );
164  }
165 
166  $next = null;
167  do {
168  $limit = min( $this->getBatchSize(), $end - $start + 1 ); // don't go pass ending ID
169  $this->output( "Doing id $start to " . ( $start + $limit - 1 ) . "...\n" );
170 
171  $entries = $src->getJournal()->getChangeEntries( $start, $limit, $next );
172  $start = $next; // start where we left off next time
173  if ( $first && !count( $entries ) ) {
174  return false; // nothing to do
175  }
176  $first = false;
177 
178  $lastPosInBatch = 0;
179  $pathsInBatch = []; // changed paths
180  foreach ( $entries as $entry ) {
181  if ( $entry['op'] !== 'null' ) { // null ops are just for reference
182  $pathsInBatch[$entry['path']] = 1; // remove duplicates
183  }
184  $lastPosInBatch = $entry['id'];
185  }
186 
187  $status = $this->syncFileBatch( array_keys( $pathsInBatch ), $src, $dst );
188  if ( $status->isOK() ) {
189  $lastOKPos = max( $lastOKPos, $lastPosInBatch );
190  $callback( $lastOKPos ); // update position file
191  } else {
192  $this->error( print_r( $status->getErrorsArray(), true ) );
193  break; // no gaps; everything up to $lastPos must be OK
194  }
195 
196  if ( !$start ) {
197  $this->output( "End of journal entries.\n" );
198  }
199  } while ( $start && $start <= $end );
200 
201  return $lastOKPos;
202  }
203 
212  protected function syncFileBatch( array $paths, FileBackend $src, FileBackend $dst ) {
213  $status = Status::newGood();
214  if ( !count( $paths ) ) {
215  return $status; // nothing to do
216  }
217 
218  // Source: convert internal backend names (FileBackendMultiWrite) to the public one
219  $sPaths = $this->replaceNamePaths( $paths, $src );
220  // Destination: get corresponding path name
221  $dPaths = $this->replaceNamePaths( $paths, $dst );
222 
223  // Lock the live backend paths from modification
224  $sLock = $src->getScopedFileLocks( $sPaths, LockManager::LOCK_UW, $status );
225  $eLock = $dst->getScopedFileLocks( $dPaths, LockManager::LOCK_EX, $status );
226  if ( !$status->isOK() ) {
227  return $status;
228  }
229 
230  $src->preloadFileStat( [ 'srcs' => $sPaths, 'latest' => 1 ] );
231  $dst->preloadFileStat( [ 'srcs' => $dPaths, 'latest' => 1 ] );
232 
233  $ops = [];
234  $fsFiles = [];
235  foreach ( $sPaths as $i => $sPath ) {
236  $dPath = $dPaths[$i]; // destination
237  $sExists = $src->fileExists( [ 'src' => $sPath, 'latest' => 1 ] );
238  if ( $sExists === true ) { // exists in source
239  if ( $this->filesAreSame( $src, $dst, $sPath, $dPath ) ) {
240  continue; // avoid local copies for non-FS backends
241  }
242  // Note: getLocalReference() is fast for FS backends
243  $fsFile = $src->getLocalReference( [ 'src' => $sPath, 'latest' => 1 ] );
244  if ( !$fsFile ) {
245  $this->error( "Unable to sync '$dPath': could not get local copy." );
246  $status->fatal( 'backend-fail-internal', $src->getName() );
247 
248  return $status;
249  }
250  $fsFiles[] = $fsFile; // keep TempFSFile objects alive as needed
251  // Note: prepare() is usually fast for key/value backends
252  $status->merge( $dst->prepare( [
253  'dir' => dirname( $dPath ), 'bypassReadOnly' => 1 ] ) );
254  if ( !$status->isOK() ) {
255  return $status;
256  }
257  $ops[] = [ 'op' => 'store',
258  'src' => $fsFile->getPath(), 'dst' => $dPath, 'overwrite' => 1 ];
259  } elseif ( $sExists === false ) { // does not exist in source
260  $ops[] = [ 'op' => 'delete', 'src' => $dPath, 'ignoreMissingSource' => 1 ];
261  } else {
262  $this->error( "Unable to sync '$dPath': could not stat file." );
263  $status->fatal( 'backend-fail-internal', $src->getName() );
264 
265  return $status;
266  }
267  }
268 
269  $t_start = microtime( true );
270  $status = $dst->doQuickOperations( $ops, [ 'bypassReadOnly' => 1 ] );
271  if ( !$status->isOK() ) {
272  sleep( 10 ); // wait and retry copy again
273  $status = $dst->doQuickOperations( $ops, [ 'bypassReadOnly' => 1 ] );
274  }
275  $elapsed_ms = floor( ( microtime( true ) - $t_start ) * 1000 );
276  if ( $status->isOK() && $this->getOption( 'verbose' ) ) {
277  $this->output( "Synchronized these file(s) [{$elapsed_ms}ms]:\n" .
278  implode( "\n", $dPaths ) . "\n" );
279  }
280 
281  return $status;
282  }
283 
291  protected function replaceNamePaths( $paths, FileBackend $backend ) {
292  return preg_replace(
293  '!^mwstore://([^/]+)!',
294  StringUtils::escapeRegexReplacement( "mwstore://" . $backend->getName() ),
295  $paths // string or array
296  );
297  }
298 
299  protected function filesAreSame( FileBackend $src, FileBackend $dst, $sPath, $dPath ) {
300  return (
301  ( $src->getFileSize( [ 'src' => $sPath ] )
302  === $dst->getFileSize( [ 'src' => $dPath ] ) // short-circuit
303  ) && ( $src->getFileSha1Base36( [ 'src' => $sPath ] )
304  === $dst->getFileSha1Base36( [ 'src' => $dPath ] )
305  )
306  );
307  }
308 }
309 
310 $maintClass = SyncFileBackend::class;
311 require_once RUN_MAINTENANCE_IF_MAIN;
const RUN_MAINTENANCE_IF_MAIN
Definition: Maintenance.php:39
error( $err, $die=0)
Throw an error to the user.
getOption( $name, $default=null)
Get an option, or return the default.
Maintenance script that syncs one file backend to another based on the journal of later...
static escapeRegexReplacement( $string)
Escape a string to make it suitable for inclusion in a preg_replace() replacement parameter...
Abstract maintenance class for quickly writing and churning out maintenance scripts with minimal effo...
Definition: Maintenance.php:82
getName()
Get the unique backend name.
getScopedFileLocks(array $paths, $type, StatusValue $status, $timeout=0)
Lock the files at the given storage paths in the backend.
setBatchSize( $s=0)
Set the batch size.
hasOption( $name)
Checks to see if a particular option exists.
syncBackends(FileBackend $src, FileBackend $dst, $start, $end, Closure $callback)
Sync $dst backend to $src backend based on the $src logs given after $start.
const LOCK_UW
Definition: LockManager.php:69
const LOCK_EX
Definition: LockManager.php:70
static newGood( $value=null)
Factory function for good results.
Definition: StatusValue.php:81
addDescription( $text)
Set the description text.
fileExists(array $params)
Check if a file exists at a storage path in the backend.
getFileSha1Base36(array $params)
Get a SHA-1 hash of the content of the file at a storage path in the backend.
getFileSize(array $params)
Get the size (bytes) of a file at a storage path in the backend.
output( $out, $channel=null)
Throw some output to the user.
doQuickOperations(array $ops, array $opts=[])
Perform a set of independent file operations on some files.
prepare(array $params)
Prepare a storage directory for usage.
getJournal()
Get the file journal object for this backend.
getBatchSize()
Returns batch size.
Base class for all file backend classes (including multi-write backends).
Definition: FileBackend.php:94
syncFileBatch(array $paths, FileBackend $src, FileBackend $dst)
Sync particular files of backend $src to the corresponding $dst backend files.
addOption( $name, $description, $required=false, $withArg=false, $shortName=false, $multiOccurrence=false)
Add a parameter to the script.
$maintClass
fatalError( $msg, $exitCode=1)
Output a message and terminate the current script.
preloadFileStat(array $params)
Preload file stat information (concurrently if possible) into in-process cache.
getLocalReference(array $params)
Returns a file system file, identical in content to the file at a storage path.
filesAreSame(FileBackend $src, FileBackend $dst, $sPath, $dPath)
return true
Definition: router.php:82
replaceNamePaths( $paths, FileBackend $backend)
Substitute the backend name of storage paths with that of a given one.