MediumSpecificBagOStuff.php

Go to the documentation of this file.

<?php
use Wikimedia\WaitConditionLoop;
 
abstract class MediumSpecificBagOStuff extends BagOStuff {
    protected $locks = [];
    protected $lastError = self::ERR_NONE;
    protected $keyspace = 'local';
    protected $syncTimeout;
    protected $segmentationSize;
    protected $segmentedValueMaxSize;
 
    private $duplicateKeyLookups = [];
    private $reportDupes = false;
    private $dupeTrackScheduled = false;
 
    protected $busyCallbacks = [];
 
    protected $preparedValues = [];
 
    const SEGMENT_COMPONENT = 'segment';
 
    public function __construct( array $params = [] ) {
        parent::__construct( $params );
 
        if ( isset( $params['keyspace'] ) ) {
            $this->keyspace = $params['keyspace'];
        }
 
        if ( !empty( $params['reportDupes'] ) && is_callable( $this->asyncHandler ) ) {
            $this->reportDupes = true;
        }
 
        $this->syncTimeout = $params['syncTimeout'] ?? 3;
        $this->segmentationSize = $params['segmentationSize'] ?? 8388608; // 8MiB
        $this->segmentedValueMaxSize = $params['segmentedValueMaxSize'] ?? 67108864; // 64MiB
    }
 
    public function get( $key, $flags = 0 ) {
        $this->trackDuplicateKeys( $key );
 
        return $this->resolveSegments( $key, $this->doGet( $key, $flags ) );
    }
 
    private function trackDuplicateKeys( $key ) {
        if ( !$this->reportDupes ) {
            return;
        }
 
        if ( !isset( $this->duplicateKeyLookups[$key] ) ) {
            // Track that we have seen this key. This N-1 counting style allows
            // easy filtering with array_filter() later.
            $this->duplicateKeyLookups[$key] = 0;
        } else {
            $this->duplicateKeyLookups[$key] += 1;
 
            if ( $this->dupeTrackScheduled === false ) {
                $this->dupeTrackScheduled = true;
                // Schedule a callback that logs keys processed more than once by get().
                call_user_func( $this->asyncHandler, function () {
                    $dups = array_filter( $this->duplicateKeyLookups );
                    foreach ( $dups as $key => $count ) {
                        $this->logger->warning(
                            'Duplicate get(): "{key}" fetched {count} times',
                            // Count is N-1 of the actual lookup count
                            [ 'key' => $key, 'count' => $count + 1, ]
                        );
                    }
                } );
            }
        }
    }
 
    abstract protected function doGet( $key, $flags = 0, &$casToken = null );
 
    public function set( $key, $value, $exptime = 0, $flags = 0 ) {
        list( $entry, $usable ) = $this->makeValueOrSegmentList( $key, $value, $exptime, $flags );
        // Only when all segments (if any) are stored should the main key be changed
        return $usable ? $this->doSet( $key, $entry, $exptime, $flags ) : false;
    }
 
    abstract protected function doSet( $key, $value, $exptime = 0, $flags = 0 );
 
    public function delete( $key, $flags = 0 ) {
        if ( !$this->fieldHasFlags( $flags, self::WRITE_PRUNE_SEGMENTS ) ) {
            return $this->doDelete( $key, $flags );
        }
 
        $mainValue = $this->doGet( $key, self::READ_LATEST );
        if ( !$this->doDelete( $key, $flags ) ) {
            return false;
        }
 
        if ( !SerializedValueContainer::isSegmented( $mainValue ) ) {
            return true; // no segments to delete
        }
 
        $orderedKeys = array_map(
            function ( $segmentHash ) use ( $key ) {
                return $this->makeGlobalKey( self::SEGMENT_COMPONENT, $key, $segmentHash );
            },
            $mainValue->{SerializedValueContainer::SEGMENTED_HASHES}
        );
 
        return $this->deleteMulti( $orderedKeys, $flags & ~self::WRITE_PRUNE_SEGMENTS );
    }
 
    abstract protected function doDelete( $key, $flags = 0 );
 
    public function add( $key, $value, $exptime = 0, $flags = 0 ) {
        list( $entry, $usable ) = $this->makeValueOrSegmentList( $key, $value, $exptime, $flags );
        // Only when all segments (if any) are stored should the main key be changed
        return $usable ? $this->doAdd( $key, $entry, $exptime, $flags ) : false;
    }
 
    abstract protected function doAdd( $key, $value, $exptime = 0, $flags = 0 );
 
    public function merge( $key, callable $callback, $exptime = 0, $attempts = 10, $flags = 0 ) {
        return $this->mergeViaCas( $key, $callback, $exptime, $attempts, $flags );
    }
 
    final protected function mergeViaCas( $key, callable $callback, $exptime, $attempts, $flags ) {
        $attemptsLeft = $attempts;
        do {
            $token = null; // passed by reference
            // Get the old value and CAS token from cache
            $this->clearLastError();
            $currentValue = $this->resolveSegments(
                $key,
                $this->doGet( $key, $flags, $token )
            );
            if ( $this->getLastError() ) {
                // Don't spam slow retries due to network problems (retry only on races)
                $this->logger->warning(
                    __METHOD__ . ' failed due to read I/O error on get() for {key}.',
                    [ 'key' => $key ]
                );
                $success = false;
                break;
            }
 
            // Derive the new value from the old value
            $value = call_user_func( $callback, $this, $key, $currentValue, $exptime );
            $keyWasNonexistant = ( $currentValue === false );
            $valueMatchesOldValue = ( $value === $currentValue );
            unset( $currentValue ); // free RAM in case the value is large
 
            $this->clearLastError();
            if ( $value === false || $exptime < 0 ) {
                $success = true; // do nothing
            } elseif ( $valueMatchesOldValue && $attemptsLeft !== $attempts ) {
                $success = true; // recently set by another thread to the same value
            } elseif ( $keyWasNonexistant ) {
                // Try to create the key, failing if it gets created in the meantime
                $success = $this->add( $key, $value, $exptime, $flags );
            } else {
                // Try to update the key, failing if it gets changed in the meantime
                $success = $this->cas( $token, $key, $value, $exptime, $flags );
            }
            if ( $this->getLastError() ) {
                // Don't spam slow retries due to network problems (retry only on races)
                $this->logger->warning(
                    __METHOD__ . ' failed due to write I/O error for {key}.',
                    [ 'key' => $key ]
                );
                $success = false;
                break;
            }
 
        } while ( !$success && --$attemptsLeft );
 
        return $success;
    }
 
    protected function cas( $casToken, $key, $value, $exptime = 0, $flags = 0 ) {
        if ( $casToken === null ) {
            $this->logger->warning(
                __METHOD__ . ' got empty CAS token for {key}.',
                [ 'key' => $key ]
            );
 
            return false; // caller may have meant to use add()?
        }
 
        list( $entry, $usable ) = $this->makeValueOrSegmentList( $key, $value, $exptime, $flags );
        // Only when all segments (if any) are stored should the main key be changed
        return $usable ? $this->doCas( $casToken, $key, $entry, $exptime, $flags ) : false;
    }
 
    protected function doCas( $casToken, $key, $value, $exptime = 0, $flags = 0 ) {
        // @TODO: the lock() call assumes that all other relavent sets() use one
        if ( !$this->lock( $key, 0 ) ) {
            return false; // non-blocking
        }
 
        $curCasToken = null; // passed by reference
        $this->clearLastError();
        $this->doGet( $key, self::READ_LATEST, $curCasToken );
        if ( is_object( $curCasToken ) ) {
            // Using === does not work with objects since it checks for instance identity
            throw new UnexpectedValueException( "CAS token cannot be an object" );
        }
        if ( $this->getLastError() ) {
            // Fail if the old CAS token could not be read
            $success = false;
            $this->logger->warning(
                __METHOD__ . ' failed due to write I/O error for {key}.',
                [ 'key' => $key ]
            );
        } elseif ( $casToken === $curCasToken ) {
            $success = $this->doSet( $key, $value, $exptime, $flags );
        } else {
            $success = false; // mismatched or failed
            $this->logger->info(
                __METHOD__ . ' failed due to race condition for {key}.',
                [ 'key' => $key ]
            );
        }
 
        $this->unlock( $key );
 
        return $success;
    }
 
    public function changeTTL( $key, $exptime = 0, $flags = 0 ) {
        return $this->doChangeTTL( $key, $exptime, $flags );
    }
 
    protected function doChangeTTL( $key, $exptime, $flags ) {
        if ( !$this->lock( $key, 0 ) ) {
            return false;
        }
 
        $expiry = $this->getExpirationAsTimestamp( $exptime );
        $delete = ( $expiry != self::TTL_INDEFINITE && $expiry < $this->getCurrentTime() );
 
        // Use doGet() to avoid having to trigger resolveSegments()
        $blob = $this->doGet( $key, self::READ_LATEST );
        if ( $blob ) {
            if ( $delete ) {
                $ok = $this->doDelete( $key, $flags );
            } else {
                $ok = $this->doSet( $key, $blob, $exptime, $flags );
            }
        } else {
            $ok = false;
        }
 
        $this->unlock( $key );
 
        return $ok;
    }
 
    public function lock( $key, $timeout = 6, $expiry = 6, $rclass = '' ) {
        // Avoid deadlocks and allow lock reentry if specified
        if ( isset( $this->locks[$key] ) ) {
            if ( $rclass != '' && $this->locks[$key]['class'] === $rclass ) {
                ++$this->locks[$key]['depth'];
                return true;
            } else {
                return false;
            }
        }
 
        $fname = __METHOD__;
        $expiry = min( $expiry ?: INF, self::TTL_DAY );
        $loop = new WaitConditionLoop(
            function () use ( $key, $expiry, $fname ) {
                $this->clearLastError();
                if ( $this->add( "{$key}:lock", 1, $expiry ) ) {
                    return WaitConditionLoop::CONDITION_REACHED; // locked!
                } elseif ( $this->getLastError() ) {
                    $this->logger->warning(
                        $fname . ' failed due to I/O error for {key}.',
                        [ 'key' => $key ]
                    );
 
                    return WaitConditionLoop::CONDITION_ABORTED; // network partition?
                }
 
                return WaitConditionLoop::CONDITION_CONTINUE;
            },
            $timeout
        );
 
        $code = $loop->invoke();
        $locked = ( $code === $loop::CONDITION_REACHED );
        if ( $locked ) {
            $this->locks[$key] = [ 'class' => $rclass, 'depth' => 1 ];
        } elseif ( $code === $loop::CONDITION_TIMED_OUT ) {
            $this->logger->warning(
                "$fname failed due to timeout for {key}.",
                [ 'key' => $key, 'timeout' => $timeout ]
            );
        }
 
        return $locked;
    }
 
    public function unlock( $key ) {
        if ( !isset( $this->locks[$key] ) ) {
            return false;
        }
 
        if ( --$this->locks[$key]['depth'] <= 0 ) {
            unset( $this->locks[$key] );
 
            $ok = $this->doDelete( "{$key}:lock" );
            if ( !$ok ) {
                $this->logger->warning(
                    __METHOD__ . ' failed to release lock for {key}.',
                    [ 'key' => $key ]
                );
            }
 
            return $ok;
        }
 
        return true;
    }
 
    public function deleteObjectsExpiringBefore(
        $timestamp,
        callable $progress = null,
        $limit = INF
    ) {
        return false;
    }
 
    public function getMulti( array $keys, $flags = 0 ) {
        $foundByKey = $this->doGetMulti( $keys, $flags );
 
        $res = [];
        foreach ( $keys as $key ) {
            // Resolve one blob at a time (avoids too much I/O at once)
            if ( array_key_exists( $key, $foundByKey ) ) {
                // A value should not appear in the key if a segment is missing
                $value = $this->resolveSegments( $key, $foundByKey[$key] );
                if ( $value !== false ) {
                    $res[$key] = $value;
                }
            }
        }
 
        return $res;
    }
 
    protected function doGetMulti( array $keys, $flags = 0 ) {
        $res = [];
        foreach ( $keys as $key ) {
            $val = $this->doGet( $key, $flags );
            if ( $val !== false ) {
                $res[$key] = $val;
            }
        }
 
        return $res;
    }
 
    public function setMulti( array $data, $exptime = 0, $flags = 0 ) {
        if ( $this->fieldHasFlags( $flags, self::WRITE_ALLOW_SEGMENTS ) ) {
            throw new InvalidArgumentException( __METHOD__ . ' got WRITE_ALLOW_SEGMENTS' );
        }
 
        return $this->doSetMulti( $data, $exptime, $flags );
    }
 
    protected function doSetMulti( array $data, $exptime = 0, $flags = 0 ) {
        $res = true;
        foreach ( $data as $key => $value ) {
            $res = $this->doSet( $key, $value, $exptime, $flags ) && $res;
        }
 
        return $res;
    }
 
    public function deleteMulti( array $keys, $flags = 0 ) {
        if ( $this->fieldHasFlags( $flags, self::WRITE_PRUNE_SEGMENTS ) ) {
            throw new InvalidArgumentException( __METHOD__ . ' got WRITE_PRUNE_SEGMENTS' );
        }
 
        return $this->doDeleteMulti( $keys, $flags );
    }
 
    protected function doDeleteMulti( array $keys, $flags = 0 ) {
        $res = true;
        foreach ( $keys as $key ) {
            $res = $this->doDelete( $key, $flags ) && $res;
        }
        return $res;
    }
 
    public function changeTTLMulti( array $keys, $exptime, $flags = 0 ) {
        $res = true;
        foreach ( $keys as $key ) {
            $res = $this->doChangeTTL( $key, $exptime, $flags ) && $res;
        }
 
        return $res;
    }
 
    public function incrWithInit( $key, $exptime, $value = 1, $init = null, $flags = 0 ) {
        $init = is_int( $init ) ? $init : $value;
        $this->clearLastError();
        $newValue = $this->incr( $key, $value, $flags );
        if ( $newValue === false && !$this->getLastError() ) {
            // No key set; initialize
            $newValue = $this->add( $key, (int)$init, $exptime, $flags ) ? $init : false;
            if ( $newValue === false && !$this->getLastError() ) {
                // Raced out initializing; increment
                $newValue = $this->incr( $key, $value, $flags );
            }
        }
 
        return $newValue;
    }
 
    final protected function resolveSegments( $key, $mainValue ) {
        if ( SerializedValueContainer::isUnified( $mainValue ) ) {
            return $this->unserialize( $mainValue->{SerializedValueContainer::UNIFIED_DATA} );
        }
 
        if ( SerializedValueContainer::isSegmented( $mainValue ) ) {
            $orderedKeys = array_map(
                function ( $segmentHash ) use ( $key ) {
                    return $this->makeGlobalKey( self::SEGMENT_COMPONENT, $key, $segmentHash );
                },
                $mainValue->{SerializedValueContainer::SEGMENTED_HASHES}
            );
 
            $segmentsByKey = $this->doGetMulti( $orderedKeys );
 
            $parts = [];
            foreach ( $orderedKeys as $segmentKey ) {
                if ( isset( $segmentsByKey[$segmentKey] ) ) {
                    $parts[] = $segmentsByKey[$segmentKey];
                } else {
                    return false; // missing segment
                }
            }
 
            return $this->unserialize( implode( '', $parts ) );
        }
 
        return $mainValue;
    }
 
    public function getLastError() {
        return $this->lastError;
    }
 
    public function clearLastError() {
        $this->lastError = self::ERR_NONE;
    }
 
    protected function setLastError( $err ) {
        $this->lastError = $err;
    }
 
    final public function addBusyCallback( callable $workCallback ) {
        $this->busyCallbacks[] = $workCallback;
    }
 
    final protected function makeValueOrSegmentList( $key, $value, $exptime, $flags ) {
        $entry = $value;
        $usable = true;
 
        if (
            $this->fieldHasFlags( $flags, self::WRITE_ALLOW_SEGMENTS ) &&
            !is_int( $value ) && // avoid breaking incr()/decr()
            is_finite( $this->segmentationSize )
        ) {
            $segmentSize = $this->segmentationSize;
            $maxTotalSize = $this->segmentedValueMaxSize;
 
            $serialized = $this->getSerialized( $value, $key );
            $size = strlen( $serialized );
            if ( $size > $maxTotalSize ) {
                $this->logger->warning(
                    "Value for {key} exceeds $maxTotalSize bytes; cannot segment.",
                    [ 'key' => $key ]
                );
            } elseif ( $size <= $segmentSize ) {
                // The serialized value was already computed, so just use it inline
                $entry = SerializedValueContainer::newUnified( $serialized );
            } else {
                // Split the serialized value into chunks and store them at different keys
                $chunksByKey = [];
                $segmentHashes = [];
                $count = intdiv( $size, $segmentSize ) + ( ( $size % $segmentSize ) ? 1 : 0 );
                for ( $i = 0; $i < $count; ++$i ) {
                    $segment = substr( $serialized, $i * $segmentSize, $segmentSize );
                    $hash = sha1( $segment );
                    $chunkKey = $this->makeGlobalKey( self::SEGMENT_COMPONENT, $key, $hash );
                    $chunksByKey[$chunkKey] = $segment;
                    $segmentHashes[] = $hash;
                }
                $flags &= ~self::WRITE_ALLOW_SEGMENTS; // sanity
                $usable = $this->setMulti( $chunksByKey, $exptime, $flags );
                $entry = SerializedValueContainer::newSegmented( $segmentHashes );
            }
        }
 
        return [ $entry, $usable ];
    }
 
    final protected function isRelativeExpiration( $exptime ) {
        return ( $exptime !== self::TTL_INDEFINITE && $exptime < ( 10 * self::TTL_YEAR ) );
    }
 
    final protected function getExpirationAsTimestamp( $exptime ) {
        if ( $exptime == self::TTL_INDEFINITE ) {
            return $exptime;
        }
 
        return $this->isRelativeExpiration( $exptime )
            ? intval( $this->getCurrentTime() + $exptime )
            : $exptime;
    }
 
    final protected function getExpirationAsTTL( $exptime ) {
        if ( $exptime == self::TTL_INDEFINITE ) {
            return $exptime;
        }
 
        return $this->isRelativeExpiration( $exptime )
            ? $exptime
            : (int)max( $exptime - $this->getCurrentTime(), 1 );
    }
 
    final protected function isInteger( $value ) {
        if ( is_int( $value ) ) {
            return true;
        } elseif ( !is_string( $value ) ) {
            return false;
        }
 
        $integer = (int)$value;
 
        return ( $value === (string)$integer );
    }
 
    public function makeKeyInternal( $keyspace, $args ) {
        $key = $keyspace;
        foreach ( $args as $arg ) {
            $key .= ':' . str_replace( ':', '%3A', $arg );
        }
        return strtr( $key, ' ', '_' );
    }
 
    public function makeGlobalKey( $class, ...$components ) {
        return $this->makeKeyInternal( 'global', func_get_args() );
    }
 
    public function makeKey( $class, ...$components ) {
        return $this->makeKeyInternal( $this->keyspace, func_get_args() );
    }
 
    public function getQoS( $flag ) {
        return $this->attrMap[$flag] ?? self::QOS_UNKNOWN;
    }
 
    public function getSegmentationSize() {
        return $this->segmentationSize;
    }
 
    public function getSegmentedValueMaxSize() {
        return $this->segmentedValueMaxSize;
    }
 
    public function setNewPreparedValues( array $valuesByKey ) {
        $this->preparedValues = [];
 
        $sizes = [];
        foreach ( $valuesByKey as $key => $value ) {
            if ( $value === false ) {
                $sizes[] = null; // not storable, don't bother
                continue;
            }
 
            $serialized = $this->serialize( $value );
            $sizes[] = ( $serialized !== false ) ? strlen( $serialized ) : null;
 
            $this->preparedValues[$key] = [ $value, $serialized ];
        }
 
        return $sizes;
    }
 
    protected function getSerialized( $value, $key ) {
        // Reuse any available prepared (serialized) value
        if ( array_key_exists( $key, $this->preparedValues ) ) {
            list( $prepValue, $prepSerialized ) = $this->preparedValues[$key];
            // Normally, this comparison should only take a few microseconds to confirm a match.
            // Using "===" on variables of different types is always fast. It is also fast for
            // variables of matching type int, float, bool, null, and object. Lastly, it is fast
            // for comparing arrays/strings if they are copy-on-write references, which should be
            // the case at this point, assuming prepareValues() was called correctly.
            if ( $prepValue === $value ) {
                unset( $this->preparedValues[$key] );
 
                return $prepSerialized;
            }
        }
 
        return $this->serialize( $value );
    }
 
    protected function guessSerialValueSize( $value, $depth = 0, &$loops = 0 ) {
        // Include serialization format overhead estimates roughly based on serialize(),
        // without counting . Also, int/float variables use the largest case
        // byte size for numbers of that type; this avoids CPU overhead for large arrays.
        switch ( gettype( $value ) ) {
            case 'string':
                // E.g. "<type><delim1><quote><value><quote><delim2>"
                return strlen( $value ) + 5;
            case 'integer':
                // E.g. "<type><delim1><sign><2^63><delim2>";
                // ceil(log10 (2^63)) = 19
                return 23;
            case 'double':
                // E.g. "<type><delim1><sign><2^52><esign><2^10><delim2>"
                // ceil(log10 (2^52)) = 16 and ceil(log10 (2^10)) = 4
                return 25;
            case 'boolean':
                // E.g. "true" becomes "1" and "false" is not storable
                return $value ? 1 : null;
            case 'NULL':
                return 1; // "\0"
            case 'array':
            case 'object':
                // Give up up and guess if there is too much depth
                if ( $depth >= 5 && $loops >= 256 ) {
                    return 1024;
                }
 
                ++$loops;
                // E.g. "<type><delim1><brace><<Kn><Vn> for all n><brace><delim2>"
                $size = 5;
                // Note that casting to an array includes private object members
                foreach ( (array)$value as $k => $v ) {
                    // Inline the recursive result here for performance
                    $size += is_string( $k ) ? ( strlen( $k ) + 5 ) : 23;
                    $size += $this->guessSerialValueSize( $v, $depth + 1, $loops );
                }
 
                return $size;
            default:
                return null; // invalid
        }
    }
 
    protected function serialize( $value ) {
        return is_int( $value ) ? $value : serialize( $value );
    }
 
    protected function unserialize( $value ) {
        return $this->isInteger( $value ) ? (int)$value : unserialize( $value );
    }
 
    protected function debug( $text ) {
        if ( $this->debugMode ) {
            $this->logger->debug( "{class} debug: $text", [ 'class' => static::class ] );
        }
    }
}