com.bigdata.service.CachingMetadataIndex Maven / Gradle / Ivy
/*
Copyright (C) SYSTAP, LLC DBA Blazegraph 2006-2016. All rights reserved.
Contact:
SYSTAP, LLC DBA Blazegraph
2501 Calvert ST NW #106
Washington, DC 20008
[email protected]
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation; version 2 of the License.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program; if not, write to the Free Software
Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
*/
/*
* Created on Oct 3, 2008
*/
package com.bigdata.service;
import java.util.concurrent.locks.Lock;
import java.util.concurrent.locks.ReadWriteLock;
import java.util.concurrent.locks.ReentrantReadWriteLock;
import com.bigdata.btree.IRangeQuery;
import com.bigdata.btree.ITupleIterator;
import com.bigdata.btree.UnisolatedReadWriteIndex;
import com.bigdata.journal.ITx;
import com.bigdata.journal.TimestampUtility;
import com.bigdata.mdi.MetadataIndex.MetadataIndexMetadata;
import com.bigdata.mdi.PartitionLocator;
import cutthecrap.utils.striterators.IFilter;
/**
* Implementation caches all locators and then updates them on demand as stale
* locators are discovered.
*
* @author Bryan Thompson
* @version $Id$
*/
public class CachingMetadataIndex extends CacheOnceMetadataIndex {
/**
* The delegate from which we refresh our local copy when we see stale
* locators.
*/
private final NoCacheMetadataIndexView delegate;
/**
* Note: This class must impose synchronization on access to the B+Tree
* caching the locators. That synchronization is required since the class
* will re-fetch locators on demand in
* {@link #staleLocator(PartitionLocator)}. Since the fetched locators will
* be written onto the B+Tree cache and since the B+Tree is NOT thread-safe
* if there is a writer, then we must synchronized access to that B+Tree for
* the readers as well as the writer. We do this using a
* {@link ReadWriteLock}, which permits higher concurrency for readers.
* This should work out well as reads should far outway writes (stale
* locators are relatively rare).
*/
private final ReadWriteLock readWriteLock = new ReentrantReadWriteLock();
/**
* Cache the index partition metadata in the client.
*
* @param name
* The name of the scale-out index.
*
* @return The cached partition metadata -or- null iff there
* is no such scale-out index.
*/
public CachingMetadataIndex(final AbstractScaleOutFederation fed,
final String name, final long timestamp,
final MetadataIndexMetadata mdmd) {
super(fed, name, timestamp, mdmd);
this.delegate = new NoCacheMetadataIndexView(fed, name, timestamp, mdmd);
}
/**
* Re-fetches the locator(s).
*/
public void staleLocator(final PartitionLocator locator) {
if (locator == null)
throw new IllegalArgumentException();
if (timestamp != ITx.UNISOLATED && timestamp != ITx.READ_COMMITTED) {
/*
* Stale locator exceptions should not be thrown for these views.
*/
throw new RuntimeException(
"Stale locator, but views should be consistent? timestamp="
+ TimestampUtility.toString(timestamp));
}
if(log.isInfoEnabled())
log.info(locator.toString());
final Lock lock = readWriteLock.writeLock();
lock.lock();
try {
/*
* Now that we have a write lock, update the local cache for all
* locators known to the remote index that are spanned by the
* key-range partition associated with the stale locator.
*/
cacheLocators(//
locator.getLeftSeparatorKey(), // fromKey
locator.getRightSeparatorKey() // toKey
);
} finally {
lock.unlock();
}
}
public PartitionLocator get(final byte[] key) {
final Lock lock = readWriteLock.readLock();
lock.lock();
try {
return super.get(key);
} finally {
lock.unlock();
}
}
public PartitionLocator find(final byte[] key) {
final Lock lock = readWriteLock.readLock();
lock.lock();
try {
return super.find(key);
} finally {
lock.unlock();
}
}
public long rangeCount() {
final Lock lock = readWriteLock.readLock();
lock.lock();
try {
return delegate.rangeCount();
} finally {
lock.unlock();
}
}
public long rangeCount(final byte[] fromKey, final byte[] toKey) {
final Lock lock = readWriteLock.readLock();
lock.lock();
try {
return delegate.rangeCount(fromKey, toKey);
} finally {
lock.unlock();
}
}
public long rangeCountExact(final byte[] fromKey, final byte[] toKey) {
final Lock lock = readWriteLock.readLock();
lock.lock();
try {
return delegate.rangeCountExact(fromKey, toKey);
} finally {
lock.unlock();
}
}
public ITupleIterator rangeIterator() {
return delegate.rangeIterator(null, null);
}
public ITupleIterator rangeIterator(final byte[] fromKey, final byte[] toKey) {
return rangeIterator(fromKey, toKey, 0/* capacity */,
IRangeQuery.DEFAULT, null/*filter*/);
}
/**
* FIXME this is wrong. The {@link #delegate} must be a
* {@link UnisolatedReadWriteIndex} in order to provide correct locking for
* the iterator. The class may have to be refactored in order to permit the
* behavior to be gated by an {@link UnisolatedReadWriteIndex}.
*/
public ITupleIterator rangeIterator(final byte[] fromKey, final byte[] toKey,
final int capacity, final int flags, final IFilter filter) {
final Lock lock = readWriteLock.readLock();
lock.lock();
try {
return delegate.rangeIterator(fromKey, toKey, capacity, flags, filter);
} finally {
lock.unlock();
}
}
}
© 2015 - 2025 Weber Informatics LLC | Privacy Policy