spark-reviews mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From rxin <...@git.apache.org>
Subject [GitHub] spark pull request: [SPARK-3453] Netty-based BlockTransferService,...
Date Sun, 26 Oct 2014 19:57:15 GMT
Github user rxin commented on a diff in the pull request:

    https://github.com/apache/spark/pull/2753#discussion_r19383297
  
    --- Diff: core/src/main/scala/org/apache/spark/network/netty/NettyBlockFetcher.scala ---
    @@ -0,0 +1,92 @@
    +/*
    + * 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.spark.network.netty
    +
    +import java.nio.ByteBuffer
    +import java.util
    +
    +import org.apache.spark.{SparkConf, Logging}
    +import org.apache.spark.network.BlockFetchingListener
    +import org.apache.spark.network.netty.NettyMessages._
    +import org.apache.spark.serializer.{JavaSerializer, Serializer}
    +import org.apache.spark.network.buffer.ManagedBuffer
    +import org.apache.spark.network.client.{RpcResponseCallback, ChunkReceivedCallback, SluiceClient}
    +import org.apache.spark.storage.BlockId
    +import org.apache.spark.util.Utils
    +
    +/**
    + * Responsible for holding the state for a request for a single set of blocks. This assumes
that
    + * the chunks will be returned in the same order as requested, and that there will be
exactly
    + * one chunk per block.
    + *
    + * Upon receipt of any block, the listener will be called back. Upon failure part way
through,
    + * the listener will receive a failure callback for each outstanding block.
    + */
    +class NettyBlockFetcher(
    +    serializer: Serializer,
    +    client: SluiceClient,
    +    blockIds: Seq[String],
    +    listener: BlockFetchingListener)
    +  extends Logging {
    +
    +  require(blockIds.nonEmpty)
    +
    +  val ser = serializer.newInstance()
    +
    +  var streamHandle: ShuffleStreamHandle = _
    +
    +  val chunkCallback = new ChunkReceivedCallback {
    +    // On receipt of a chunk, pass it upwards as a block.
    +    def onSuccess(chunkIndex: Int, buffer: ManagedBuffer): Unit = Utils.logUncaughtExceptions
{
    +      listener.onBlockFetchSuccess(blockIds(chunkIndex), buffer)
    +    }
    +
    +    // On receipt of a failure, fail every block from chunkIndex onwards.
    +    def onFailure(chunkIndex: Int, e: Throwable): Unit = {
    +      blockIds.drop(chunkIndex).foreach { blockId =>
    +        listener.onBlockFetchFailure(blockId, e);
    +      }
    +    }
    +  }
    +
    +  // Send the RPC to open the given set of blocks. This will return a ShuffleStreamHandle.
    +  client.sendRpc(ser.serialize(OpenBlocks(blockIds.map(BlockId.apply))).array(),
    +    new RpcResponseCallback {
    +      override def onSuccess(response: Array[Byte]): Unit = {
    +        try {
    +          streamHandle = ser.deserialize[ShuffleStreamHandle](ByteBuffer.wrap(response))
    +          logTrace(s"Successfully opened block set: $streamHandle! Preparing to fetch
chunks.")
    +
    +          // Immediately request all chunks -- we expect that the total size of the request
is
    +          // reasonable due to higher level chunking in [[ShuffleBlockFetcherIterator]].
    +          for (i <- 0 until streamHandle.numChunks) {
    +            client.fetchChunk(streamHandle.streamId, i, chunkCallback)
    +          }
    +        } catch {
    +          case e: Exception =>
    +            logError("Failed while starting block fetches", e)
    +            blockIds.foreach(listener.onBlockFetchFailure(_, e))
    --- End diff --
    
    wrap this around a try catch block so we call all block ids even if onBlockFetchFailure
throws an exception?


---
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.
---

---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org


Mime
View raw message