[ 
https://issues.apache.org/jira/browse/HDFS-14356?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16850757#comment-16850757
 ] 

Sammi Chen commented on HDFS-14356:
-----------------------------------

Thanks [~PhiloHe] for the update. A few comments,
1. typo "amd" in  "Please refer to http://pmem.io/ amd 
https://github.com/pmem/pmdk";
2.  "It is recommended to build the project with this option if user plans to 
use SCM backed
    HDFS cache."  name the reason why it's recommended, "for better 
performance"? 
3. for log the mapper block loader, try "this.class.getClass()" in super class, 
instead of use instanceof sub class name.
4.  if (!(cacheLoader instanceof NativePmemMappableBlockLoader)) {
      return -1;
    }
    Suggest add a new check function like "isNative" in MappableBlockLoader, 
instead of use instanceof to keep FsDatasetCache kind of cache loader 
implementation nutral. Same suggestion for the NativePmemMappedBlock. You can 
add a new function, such as "getMappedAddress"
5.  Suggest change the log level to debug or trace.
   LOG.info("Get InputStream by cache address."); 
6. several typo "pmemMappedAddres" in NativePmemMappedBlock
7.  4 space indention for the second line. And double check other functions 
too. 
    NativePmemMappedBlock(long pmemMappedAddres, long length,
                        ExtendedBlockId key) {

> Implement HDFS cache on SCM with native PMDK libs
> -------------------------------------------------
>
>                 Key: HDFS-14356
>                 URL: https://issues.apache.org/jira/browse/HDFS-14356
>             Project: Hadoop HDFS
>          Issue Type: Sub-task
>          Components: caching, datanode
>            Reporter: Feilong He
>            Assignee: Feilong He
>            Priority: Major
>         Attachments: HDFS-14356.000.patch, HDFS-14356.001.patch, 
> HDFS-14356.002.patch, HDFS-14356.003.patch, HDFS-14356.004.patch, 
> HDFS-14356.005.patch, HDFS-14356.006.patch
>
>
> In this implementation, native PMDK libs are used to map HDFS blocks to SCM. 
> To use this implementation, user should build hadoop with PMDK libs by 
> specifying a build option. This implementation is only supported on linux 
> platform.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

---------------------------------------------------------------------
To unsubscribe, e-mail: hdfs-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: hdfs-issues-h...@hadoop.apache.org

Reply via email to