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 15:38:25 GMT

    [ https://issues.apache.org/jira/browse/ACCUMULO-4187?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15239450#comment-15239450
] 

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

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

    https://github.com/apache/accumulo/pull/90#discussion_r59571367
  
    --- Diff: core/src/main/java/org/apache/accumulo/core/file/FileOperations.java ---
    @@ -55,23 +56,24 @@ public static FileOperations getInstance() {
        */
     
       public abstract FileSKVIterator openReader(String file, Range range, Set<ByteSequence>
columnFamilies, boolean inclusive, FileSystem fs, Configuration conf,
    -      AccumuloConfiguration tableConf) throws IOException;
    +      RateLimiter readLimiter, AccumuloConfiguration tableConf) throws IOException;
     
       public abstract FileSKVIterator openReader(String file, Range range, Set<ByteSequence>
columnFamilies, boolean inclusive, FileSystem fs, Configuration conf,
    -      AccumuloConfiguration tableConf, BlockCache dataCache, BlockCache indexCache) throws
IOException;
    +      RateLimiter readLimiter, AccumuloConfiguration tableConf, BlockCache dataCache,
BlockCache indexCache) throws IOException;
     
       /**
        * Open a reader that fully support seeking and also enable any optimizations related
to seeking, like bloom filters.
        *
        */
     
    -  public abstract FileSKVIterator openReader(String file, boolean seekToBeginning, FileSystem
fs, Configuration conf, AccumuloConfiguration acuconf)
    -      throws IOException;
    +  public abstract FileSKVIterator openReader(String file, boolean seekToBeginning, FileSystem
fs, Configuration conf, RateLimiter readLimiter,
    --- End diff --
    
    So, I've had some thoughts on this as well. A lot of our internal passing around of AccumuloConfiguration
makes it easy to lose track of where the configuration object came from, making it harder
to ensure we're not using it outside the intended scope (for example, if it represents a TableConfiguration).
    
    In many parts of our internal code, I would prefer we grab what we need from AccumuloConfiguration,
combine it with any additional configuration appropriate for that specific context, and create
a new context-specific configuration object to pass around within that context.
    
    If we were to do something like that, what would the context granularity be here? "file
operations options", "file reading options", something else?


> 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