accumulo-notifications mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "ASF GitHub Bot (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (ACCUMULO-4187) Rate limiting of major compactions
Date Wed, 13 Apr 2016 16:28:25 GMT

    [ https://issues.apache.org/jira/browse/ACCUMULO-4187?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15239545#comment-15239545
] 

ASF GitHub Bot commented on ACCUMULO-4187:
------------------------------------------

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

    https://github.com/apache/accumulo/pull/90#discussion_r59579822
  
    --- Diff: core/src/main/java/org/apache/accumulo/core/file/streams/PositionedOutputs.java
---
    @@ -0,0 +1,55 @@
    +/*
    + * 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.accumulo.core.file.streams;
    +
    +import java.io.FilterOutputStream;
    +import java.io.IOException;
    +import java.io.OutputStream;
    +import org.apache.hadoop.fs.FSDataOutputStream;
    +
    +public class PositionedOutputs {
    +  public static PositionedOutputStream wrap(final OutputStream fout) {
    +    if (fout instanceof FSDataOutputStream) {
    +      return new PositionedOutputStream(fout) {
    +        @Override
    +        public long position() throws IOException {
    +          return ((FSDataOutputStream) fout).getPos();
    +        }
    +      };
    +    } else if (fout instanceof PositionedOutput) {
    +      return new PositionedOutputStream(fout) {
    +        @Override
    +        public long position() throws IOException {
    +          return ((PositionedOutput) fout).position();
    +        }
    +      };
    +    } else {
    +      return new PositionedOutputStream(fout) {
    +        @Override
    +        public long position() throws IOException {
    +          throw new UnsupportedOperationException("Underlying stream does not support
position()");
    +        }
    +      };
    +    }
    +  }
    +
    +  public static abstract class PositionedOutputStream extends FilterOutputStream implements
PositionedOutput {
    --- End diff --
    
    What I really want to say is "This method returns an `OutputStream` which implements `PositionedOutput`"
 But there's not really a way to say that in Java.
    
    I'm probably just being too clever with `PositionedOutput` anyways.  The underlying issue
is that `BCFile.Writer` needs an `OutputStream` implementing `DataOutput` with some facsimile
of the `getPos()` method from `FSDataOutputStream`.  `RateLimitedOutputStream` doesn't want
to care about or interfere with either of these.  But there doesn't seem to be any way to
separate concerns cleanly.  `getPos()` doesn't have an interface (like `Seekable`), and `java.io.DataOutputStream`
(and so `FSDataOutputStream`) isn't implemented in a way which readily supports subclassing/inheritance.


> Rate limiting of major compactions
> ----------------------------------
>
>                 Key: ACCUMULO-4187
>                 URL: https://issues.apache.org/jira/browse/ACCUMULO-4187
>             Project: Accumulo
>          Issue Type: Improvement
>          Components: core, tserver
>    Affects Versions: 1.8.0
>            Reporter: Shawn Walker
>            Assignee: Shawn Walker
>            Priority: Minor
>             Fix For: 1.8.0
>
>
> In discussing [ACCUMULO-4166] with Keith Turner, we decided that the underlying issue
is that major compactions can overwhelm a tablet server, rendering it nearly unresponsive.
> To address this, we should take a cue from Apache Cassandra and restrict how quickly
we perform major compactions.  Rate limiting reads and writes involved in major compactions
will directly affect the IO load caused by major compactions, and should also indirectly affect
the CPU load.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message