[
https://issues.apache.org/jira/browse/HBASE-18066?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16020583#comment-16020583
]
Zheng Hu commented on HBASE-18066:
----------------------------------
bq. My question is why we implement the function with two steps? I mean first
get the row key and then use this row key to get the row. Is it only an
implementation compromise? Or intentional?
[~Apache9], If implement the function with one step(scan backward) , then the
implementation would be similar to reverse scan which was not implemented yet
(Maybe a lot of work was needed for reverse scan at that time), so we
provided the method getClosestRowBefore to locate the region in hbase:meta, I
guess.
For compatibility, we should keep getClosestRowBefore in branch-1.x & master
branch at server side, so we should fix the bug at server side.
There are two way to fix the bug, In my opinion:
1. Just implement getClosestRowBefore by reverse scan at server side.
2. Still two steps, but we get the readpoint firstly, and then pass the
readpoint in both getting rowkey step and getting row step.
Let me have a try.
> Get with closest_row_before on "hbase:meta" can return empty Cell during
> region merge/split
> -------------------------------------------------------------------------------------------
>
> Key: HBASE-18066
> URL: https://issues.apache.org/jira/browse/HBASE-18066
> Project: HBase
> Issue Type: Bug
> Components: hbase, regionserver
> Affects Versions: 1.3.1
> Environment: Linux (16.04.2), MacOS 10.11.6.
> Standalone and distributed HBase setup.
> Reporter: Andrey Elenskiy
> Attachments: TestGetWithClosestRowBeforeWhenSplit.java
>
>
> During region split/merge there's a brief period of time where doing a "Get"
> with "closest_row_before=true" on "hbase:meta" may return empty
> "GetResponse.result.cell" field even though parent, splitA and splitB regions
> are all in "hbase:meta". Both gohbase (https://github.com/tsuna/gohbase) and
> AsyncHBase (https://github.com/OpenTSDB/asynchbase) interprets this as
> "TableDoesNotExist", which is returned to the client.
> Here's a gist that reproduces this problem:
> https://gist.github.com/Timoha/c7a236b768be9220e85e53e1ca53bf96. Note that
> you have to use older HTable client (I used 1.2.4) as current versions ignore
> `Get.setClosestRowBefore(bool)` option.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
