[
https://issues.apache.org/jira/browse/HBASE-19658?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16325525#comment-16325525
]
Anastasia Braginsky commented on HBASE-19658:
---------------------------------------------
Hey [~stack] and [~jmspaggi],
As I do not have enough data, I am generally code-reviewing test
testFlatteningToJumboCellChunkMap in file
TestCompactingToCellFlatMapMemStore.java line 815, recent version of branch-2.
Also taking into consideration that test sometimes runs successfully.
The difference in the sizes is just 68 bytes, considering that in this test
huge cells (above 2MB) are tested, I would say that just something minor can
went differently. First question, whether any machine with specific hardware is
used here for testing? Can it be 32-bit words?
Another possibility is that in-memory-flush is for some reason calculated
differently in some of the cases. The tests assumes that 1 cell will stay in
the active and 2 cells are going to be flushed-in-memory and flattened into the
pipeline. However, the difference in sizes is *exactly* the one that should be
when all 3 cells are flushed-in-memory and flattened into the pipeline. How
many region servers and stores are configured? Anything special here?
As general solution we can definitely rise the size of the written cells a
little bit more (and configure the test differently) so in any sure case 3
cells are flushed-in-memory and this is tested.
> Fix and reenable
> TestCompactingToCellFlatMapMemStore#testFlatteningToJumboCellChunkMap
> --------------------------------------------------------------------------------------
>
> Key: HBASE-19658
> URL: https://issues.apache.org/jira/browse/HBASE-19658
> Project: HBase
> Issue Type: Bug
> Components: test
> Affects Versions: 2.0.0-beta-1
> Reporter: stack
> Assignee: Anastasia Braginsky
> Fix For: 2.0.0-beta-2
>
> Attachments: HBASE-19658-V01.patch, HBASE-19658-V02.patch,
> HBASE-19658-V03.patch,
> org.apache.hadoop.hbase.regionserver.TestCompactingToCellFlatMapMemStore-output.txt
>
>
> testFlatteningToJumboCellChunkMap was disabled so could commit HBASE-19282 on
> branch-2. This test is failing reliably. Assigned to [~anastas]. This issue
> is about fixing the failing test and reenabling it in time for beta-2. Thanks
> A.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
