VladRodionov commented on code in PR #8739: URL: https://github.com/apache/hbase/pull/8739#discussion_r4251551449
########## hbase-server/src/main/java/org/apache/hadoop/hbase/io/hfile/cache/persistence/HadoopFsCachePersistenceStorage.java: ########## @@ -0,0 +1,419 @@ +/* + * 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.hadoop.hbase.io.hfile.cache.persistence; + +import java.io.IOException; +import java.io.InputStream; +import java.io.OutputStream; +import java.util.Objects; +import java.util.Optional; +import java.util.UUID; +import org.apache.hadoop.conf.Configuration; +import org.apache.hadoop.fs.FSDataOutputStream; +import org.apache.hadoop.fs.FileSystem; +import org.apache.hadoop.fs.Path; +import org.apache.yetus.audience.InterfaceAudience; + +/** + * {@link CachePersistenceStorage} implementation backed by a Hadoop {@link FileSystem}. + * <p> + * This implementation works with filesystems available through Hadoop's filesystem abstraction, + * including HDFS and the local filesystem. + * </p> + * <p> + * Every persistence storage key is represented as a file below a configured root directory. Keys + * may contain relative path components separated by {@code /}, allowing higher-level persistence + * infrastructure to organize state by component role. + * </p> + * <p> + * Persistence writes are first written to a temporary file. The temporary file is promoted to the + * committed state path only after the caller explicitly commits the write. An abandoned write + * deletes its temporary file and leaves previously committed state unchanged. + * </p> + * <p> + * Publication uses filesystem rename operations. Rename atomicity and implementation details are + * determined by the underlying Hadoop filesystem. For filesystems where rename is implemented as + * copy-and-delete, publication has the corresponding filesystem semantics. + * </p> + * <p> + * This class does not own the supplied {@link FileSystem} and does not close it. + * </p> + */ [email protected] +public class HadoopFsCachePersistenceStorage implements CachePersistenceStorage { Review Comment: On your first comment. I don't intend this API to mean “backup/snapshot to non-local storage.” The distinction here is between the component persistence format and the storage used to hold it. PersistentCacheComponent only defines how a component saves/restores its own runtime state to streams. CachePersistenceStorage decides where those streams live. The provided HadoopFsCachePersistenceStorage supports the local Hadoop filesystem (file://) as well as HDFS and other Hadoop-compatible filesystems, so preserving state on the same local persistent disk is fully supported. BucketCachePersister is actually the natural first concrete use case once BucketCache is converted to a native BucketCacheEngine. Its current behavior already persists BucketCache metadata/backing-map state so that a persistent local IO engine can be reconstructed after restart. I expect that logic to become the PersistentCacheComponent.save()/restore() implementation of BucketCacheEngine, with local filesystem storage preserving the existing behavior. Remote storage is an additional possibility enabled by the abstraction, not the required semantics. In particular, storing only BucketCache metadata remotely would not by itself make a local BucketCache portable to another machine—the underlying persistent cache data must also remain available. So I would avoid describing this API as a BucketCache backup mechanism. -- 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: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
