[
https://issues.apache.org/jira/browse/HBASE-14832?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15033722#comment-15033722
]
Anoop Sam John commented on HBASE-14832:
----------------------------------------
Looks good.
Some minor comments
writeRowExcludingCommonPrefix
Call it writeRowSkippingBytes(out, Cell, int skipBytes)
Same way for writeQualifierExcludingCommonPrefix
writeRow -> Do we need to pass the rLength when we can get it from Cell? Or
may be it is already computed in calling place and u want to avoid parsing it
again? Same for writeRowExcludingCommonPrefix
{code}
int rkCommonPrefix = 0;
1119 if (c1 instanceof ByteBufferedCell && c2 instanceof
ByteBufferedCell) {
1120 rkCommonPrefix =
ByteBufferUtils.findCommonPrefix(((ByteBufferedCell) c1).getRowByteBuffer(),
1121 ((ByteBufferedCell) c1).getRowPosition(), rLen1,
((ByteBufferedCell) c2).getRowByteBuffer(),
1122 ((ByteBufferedCell) c2).getRowPosition(), rLen2);
1123 } else {
1124 // There cannot be a case where one cell is BBCell and other is
KeyValue. This flow comes either
1125 // in flush or compactions. In flushes both cells are KV and in
case of compaction it will be either
1126 // KV or BBCell
1127 rkCommonPrefix =
ByteBufferUtils.findCommonPrefix(c1.getRowArray(), c1.getRowOffset()
{code}
Can the instance of check be limited in some util place? CellUtil?
nit:
ByteBufferUtils.toByte(in, offset + i)
Just call toByte() as we are in same class
Same way in new method findCommonPrefix
> Ensure write paths work with ByteBufferedCells in case of compaction
> --------------------------------------------------------------------
>
> Key: HBASE-14832
> URL: https://issues.apache.org/jira/browse/HBASE-14832
> Project: HBase
> Issue Type: Sub-task
> Components: regionserver, Scanners
> Affects Versions: 2.0.0
> Reporter: ramkrishna.s.vasudevan
> Assignee: ramkrishna.s.vasudevan
> Priority: Minor
> Fix For: 2.0.0
>
> Attachments: HBASE-14832.patch
>
>
> Currently any cell coming out of offheap Bucketcache while compaction does a
> copy using the getXXXArray() API since write path does not work with BBCells.
> This JIRA is aimed at changing the write path to support BBCells so that this
> copy is avoided.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
