[ 
https://issues.apache.org/jira/browse/HDFS-5053?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13767131#comment-13767131
 ] 

Colin Patrick McCabe commented on HDFS-5053:
--------------------------------------------

bq. Renamed the config option to "dfs.namenode.caching.enabled". I think one 
option makes sense here since the big toggle is the creation of the 
cachedBlocksMap, which we'll need for both LRU and path-based caching.

OK.

bq. Caching factor sounds like follow-on work

agree

bq. Refactored InvalidateBlocks to have UncacheBlocks and 
InvalidateStoredBlocks subclasses with only the relevant methods. Hopefully 
looks cleaner now.

thanks, this looks a lot nicer.

bq. I agree that having a BlockInfo in the wrong map is scary. I guess I could 
just subclass a CachedBlockInfo and do a type check? If it's any solace, we 
only add cached blocks during cache report processing, and it's always a new 
BlockInfo or something coming out of the cachedBlocksMap.

I feel like we may be able to clean this up further, possibly with generics 
(BlockMap<CachBlockInfo> versus BlockMap<BlockInfo> or something).  But it may 
be best to do this later...

bq. [misreplaced blocks, safe mode]

I think we should revisit this in HDFS-5092.  For now, ignoring them is 
probably fine.

{code}
  final String className = getClass().getSimpleName();
{code}
This could be private.

{code}
  @Override // ReplicationManager
  BlockInfo processReportedBlock(final DatanodeDescriptor dn, 
{code}
Comment is wrong
                
> NameNode should invoke DataNode APIs to coordinate caching
> ----------------------------------------------------------
>
>                 Key: HDFS-5053
>                 URL: https://issues.apache.org/jira/browse/HDFS-5053
>             Project: Hadoop HDFS
>          Issue Type: Sub-task
>          Components: datanode, namenode
>            Reporter: Colin Patrick McCabe
>            Assignee: Andrew Wang
>         Attachments: hdfs-5053-1.patch, hdfs-5053-2.patch
>
>
> The NameNode should invoke the DataNode APIs to coordinate caching.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira

Reply via email to