[
https://issues.apache.org/jira/browse/HBASE-2359?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12848584#action_12848584
]
stack commented on HBASE-2359:
------------------------------
.bq Strictly speaking for HLogKey & WALEdit we are interested in the serialized
size rather than the heap size.
Yes. Good point. HeapSize is not whats wanted at all.
.bq Another possibility--- SequenceFile.Writer has a "getLength"... then do
away with implementing any size method for HLogKey & WALEdit.
+1
This is cutting to the chase, its keeping tabs on the WAL file length that is
point of sizing WALEdit+HLogKey. We'll never want to compress WAL logs so the
comment in SF#getLength about inaccuracy when compressing does not apply..
getLength is out.getPos which is a long kept up by FSDataOutputStream.
> WALEdit doesn't implement HeapSize
> ----------------------------------
>
> Key: HBASE-2359
> URL: https://issues.apache.org/jira/browse/HBASE-2359
> Project: Hadoop HBase
> Issue Type: Bug
> Affects Versions: 0.20.3
> Reporter: Jean-Daniel Cryans
> Assignee: Kannan Muthukkaruppan
> Fix For: 0.20.4, 0.21.0
>
>
> WALEdit from HBASE-2283 defines a method heapSize() but doesn't implement
> HeapSize.
> - Make it implement the interface.
> - Add a test to TestHeapSize.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.