Mailing-List: contact dev-help@apex.incubator.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@apex.incubator.apache.org
From: ilooner <git@git.apache.org>
To: dev@apex.incubator.apache.org
Reply-To: dev@apex.incubator.apache.org
References: <git-pr-145-incubator-apex-malhar@git.apache.org>
In-Reply-To: <git-pr-145-incubator-apex-malhar@git.apache.org>
Subject: [GitHub] incubator-apex-malhar pull request: APEXMALHAR-1897 added
 managed ...
Content-Type: text/plain
Message-Id: <20160321041411.7167CDFB7B@git1-us-west.apache.org>
Date: Mon, 21 Mar 2016 04:14:11 +0000 (UTC)

Github user ilooner commented on a diff in the pull request:

    https://github.com/apache/incubator-apex-malhar/pull/145#discussion_r56781175
  
    --- Diff: library/src/main/java/com/datatorrent/lib/state/managed/ManagedTimeUnifiedStateImpl.java ---
    @@ -0,0 +1,243 @@
    +/**
    + * 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 com.datatorrent.lib.state.managed;
    +
    +import java.io.DataInputStream;
    +import java.io.DataOutputStream;
    +import java.io.IOException;
    +import java.util.concurrent.Future;
    +import java.util.concurrent.LinkedBlockingQueue;
    +
    +import org.slf4j.Logger;
    +import org.slf4j.LoggerFactory;
    +
    +import org.apache.hadoop.fs.LocatedFileStatus;
    +import org.apache.hadoop.fs.RemoteIterator;
    +
    +import com.google.common.base.Preconditions;
    +import com.google.common.collect.Queues;
    +import com.google.common.util.concurrent.Futures;
    +
    +import com.datatorrent.lib.fileaccess.FileAccess;
    +import com.datatorrent.lib.state.BucketedState;
    +import com.datatorrent.netlet.util.Slice;
    +
    +/**
    + * In this implementation of {@link ManagedState} the buckets in memory are time-buckets.
    + * <p/>
    + *
    + * <b>Difference from {@link ManagedTimeStateImpl}</b>: <br/>
    + * <ol>
    + * <li>The main buckets in {@link ManagedTimeStateImpl} are unique adhoc long ids which the user provides with the
    + * key. In this implementation the main buckets are time buckets. The user provides just the time and the time bucket is
    + * derived from it.
    + * </li>
    + * <br/>
    + *
    + * <li>In regards to the bucket data on disk, in {@link ManagedTimeStateImpl} the buckets are persisted on disk
    + * with each bucket data further grouped into time-buckets: {base_path}/{bucketId}/{time-bucket id}. <br/>
    + * In this implementation operator id is used as bucketId (on disk) and there is just one time-bucket under a
    + * particular operator id:
    + * {base_path}/{operator id}/{time bucket id}.
    + * </li>
    + * <br/>
    + *
    + * <li>In {@link ManagedTimeStateImpl} a bucket belongs to just one partition. Multiple partitions cannot write to
    + * the same bucket. <br/>
    + * In this implementation multiple partitions can be working with the same time-bucket (since time-bucket is derived
    + * from time). This works because on the disk the time-bucket data is segregated under each operator id.
    + * </li>
    + * <br/>
    + *
    + * <li>While {@link ManagedTimeStateImpl} can support dynamic partitioning by pre-allocating buckets this will not
    + * be able to support dynamic partitioning efficiently.
    + * </li>
    +
    + * </ol>
    + */
    +public class ManagedTimeUnifiedStateImpl extends AbstractManagedStateImpl implements BucketedState
    +{
    +  private final transient LinkedBlockingQueue<Long> purgedTimeBuckets = Queues.newLinkedBlockingQueue();
    +
    +  public ManagedTimeUnifiedStateImpl()
    +  {
    +    bucketsFileSystem = new TimeUnifiedBucketsFileSystem();
    +  }
    +
    +  @Override
    +  public int getNumBuckets()
    +  {
    +    return timeBucketAssigner.getNumBuckets();
    +  }
    +
    +  @Override
    +  public void put(long time, Slice key, Slice value)
    +  {
    +    long timeBucket = timeBucketAssigner.getTimeBucketFor(time);
    +    if (timeBucket == -1) {
    +      //time is expired so return null.
    +      return;
    +    }
    +    int bucketIdx = prepareBucket(timeBucket);
    +
    +    buckets[bucketIdx].put(key, timeBucket, value);
    +
    +  }
    +
    +  @SuppressWarnings("SynchronizationOnLocalVariableOrMethodParameter")
    +  @Override
    +  public Slice getSync(long time, Slice key)
    +  {
    +    long timeBucket = timeBucketAssigner.getTimeBucketFor(time);
    +    if (timeBucket == -1) {
    +      //time is expired so return expired slice.
    +      return BucketedState.EXPIRED;
    +    }
    +    int bucketIdx = prepareBucket(timeBucket);
    +    Bucket bucket = buckets[bucketIdx];
    +
    +    synchronized (bucket) {
    +      return bucket.get(key, timeBucket, Bucket.ReadSource.ALL);
    +    }
    +  }
    +
    +  @SuppressWarnings("SynchronizationOnLocalVariableOrMethodParameter")
    +  @Override
    +  public Future<Slice> getAsync(long time, Slice key)
    +  {
    +    long timeBucket = timeBucketAssigner.getTimeBucketFor(time);
    +    if (timeBucket == -1) {
    +      //time is expired so return null.
    +      return Futures.immediateFuture(BucketedState.EXPIRED);
    +    }
    +    int bucketIdx = prepareBucket(timeBucket);
    +    Bucket bucket = buckets[bucketIdx];
    +    synchronized (bucket) {
    +      Slice cachedVal = buckets[bucketIdx].get(key, timeBucket, Bucket.ReadSource.MEMORY);
    +      if (cachedVal != null) {
    +        return Futures.immediateFuture(cachedVal);
    +      }
    +      return readerService.submit(new KeyFetchTask(bucket, key, timeBucket, throwable));
    +    }
    +  }
    +
    +  @Override
    +  public void endWindow()
    +  {
    +    super.endWindow();
    +    Long purgedTimeBucket;
    +
    +    //tear down all the purged time buckets
    +    while (null != (purgedTimeBucket = purgedTimeBuckets.poll())) {
    +      int purgedTimeBucketIdx = getBucketIdx(purgedTimeBucket);
    +      if (buckets[purgedTimeBucketIdx] != null && buckets[purgedTimeBucketIdx].getBucketId() == purgedTimeBucket) {
    +        buckets[purgedTimeBucketIdx].teardown();
    --- End diff --
    
    If there are asynchronous get calls happening when a bucket is purged they may attempt to retrieve data from a closed reader. I would suggest keeping track of the asynchronous get calls and wait for them to finish before purging a time bucket.


---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---