dsmiley commented on code in PR #2548: URL: https://github.com/apache/solr/pull/2548#discussion_r1704232054
########## solr/core/src/java/org/apache/solr/update/UpdateLocks.java: ########## @@ -0,0 +1,192 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one or more + * contributor license agreements. See the NOTICE file distributed with + * this work for additional information regarding copyright ownership. + * The ASF licenses this file to You under the Apache License, Version 2.0 + * (the "License"); you may not use this file except in compliance with + * the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +package org.apache.solr.update; + +import com.carrotsearch.hppc.IntObjectHashMap; +import java.io.IOException; +import java.util.ArrayDeque; +import java.util.concurrent.TimeUnit; +import java.util.concurrent.locks.Condition; +import java.util.concurrent.locks.Lock; +import java.util.concurrent.locks.ReadWriteLock; +import java.util.concurrent.locks.ReentrantLock; +import java.util.concurrent.locks.ReentrantReadWriteLock; +import org.apache.lucene.util.BytesRef; +import org.apache.solr.common.SolrException; +import org.apache.solr.common.SolrException.ErrorCode; +import org.apache.solr.util.IOFunction; + +/** + * Locks associated with updates in connection with the {@link UpdateLog}. + * + * @lucene.internal + */ +public class UpdateLocks { + + private final long docLockTimeoutMs; + + private final ReadWriteLock blockUpdatesLock = new ReentrantReadWriteLock(true); + + // SolrCloud's first approach was a fixed size array of locks (lock striping) using an ID's hash. + // Sized too small, there was too much lock sharing; sized too big, there was memory waste. + // Here we have a Map keyed by hash and a pool of locks to re-use. Synchronization is needed 2x. + // Note: ConcurrentHashMap was also explored but HPPC came out on top, probably because + // we can use a hashcode directly as the key, and it's GC friendly (zero-allocation). + + /** Maps a ID hashcode to a lock. Synchronize to manipulate. */ + private final IntObjectHashMap<LockAndCondition> hashToLock = + new IntObjectHashMap<>(32) { + @Override + protected int hashKey(int key) { + return key; // our keys are themselves hash-codes + } + }; + + /** A pool of locks to avoid creating & GC'ing them too much. Must synchronize on hashToLock. */ + private final ArrayDeque<LockAndCondition> lockPool = new ArrayDeque<>(16); + + public UpdateLocks(long docLockTimeoutMs) { + this.docLockTimeoutMs = docLockTimeoutMs; + } + + /** + * Acquires a lock for the given doc ID, executes the function, and releases the lock. The + * provided {@link Condition} can be used for wait/notify control. The per-doc locking is needed + * indirectly due to SolrCloud's internal out-of-order versioned update design. + */ + public <R> R runWithLock(BytesRef id, IOFunction<Condition, R> function) throws IOException { + final var startTimeNanos = System.nanoTime(); + + lockForUpdate(); + try { + // note: if we didn't need a Condition, then we could have reused + // OrderedExecutor.SparseStripedLock over here, which is also a mechanism invented for + // per-doc locking. + + // hashToLock isn't concurrent, but we synchronize on it briefly twice to do cheap work + + final int hash = id.hashCode(); + final LockAndCondition lock; + // get or insert lock, increment refcount + synchronized (hashToLock) { + final int idx = hashToLock.indexOf(hash); + if (hashToLock.indexExists(idx)) { + lock = hashToLock.indexGet(idx); + assert lock.refCount >= 1; + lock.refCount++; + } else { + lock = borrowLock(); + hashToLock.indexInsert(idx, hash, lock); + } + } + + // try-finally ensuring we decrement the refCount + try { + return runWithLockInternal(id, function, lock, startTimeNanos); + } finally { + // decrement refcount, remove lock if unreferenced + synchronized (hashToLock) { + assert lock.refCount > 0; // because we incremented it + if (--lock.refCount == 0) { // typical + hashToLock.remove(hash); + returnLock(lock); + } + } + } + + } finally { + unlockForUpdate(); + } + } + + private LockAndCondition borrowLock() { + assert Thread.holdsLock(hashToLock); + if (lockPool.isEmpty()) { + return new LockAndCondition(); + } else { + return lockPool.removeLast(); + } + } + + private void returnLock(LockAndCondition lock) { + assert Thread.holdsLock(hashToLock); + if (lockPool.size() < 16) { + lockPool.add(lock); + lock.refCount = 1; // ready for next use + } + } + + private <R> R runWithLockInternal( + BytesRef id, IOFunction<Condition, R> function, LockAndCondition lock, long startTimeNanos) + throws IOException { + // Acquire the lock + try { + if (docLockTimeoutMs == 0) { + lock.lock.lockInterruptibly(); + } else { + long remainingNs = + TimeUnit.MILLISECONDS.toNanos(docLockTimeoutMs) - (System.nanoTime() - startTimeNanos); Review Comment: Shrug. The conversion time is trivial. I prefer to save the metric as it was provided, which commonly is Milliseconds, but sometimes isn't, admittedly. Maybe I'll change it. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: issues-unsubscr...@solr.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org --------------------------------------------------------------------- To unsubscribe, e-mail: issues-unsubscr...@solr.apache.org For additional commands, e-mail: issues-h...@solr.apache.org
