[
https://issues.apache.org/jira/browse/HBASE-3857?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13073880#comment-13073880
]
Mikhail Bautin commented on HBASE-3857:
---------------------------------------
@Ted:
Yes, you are right, TestReplication also failed with the patch applied while it
did not fail without the patch. The failure message was as follows:
java.lang.AssertionError: Waited too much time for queueFailover replication
Looking at my archive of unit test results for the trunk, TestReplication
failure shows up in 67 cases out of 291 total runs of the test suite, so I
suspect it is a highly problematic unit test. I will look into this a bit more.
However, my question is the following: with the extremely agile development
process and a non-trivial number of routine unit test failures in the trunk,
what is the accepted approach of testing new patches? How do people select the
right revision to test their patch against? Is it always the trunk, and are the
existing unit test failures in the trunk ignored (which is bad because this may
mask new bugs), or is there a different approach?
> Change the HFile Format
> -----------------------
>
> Key: HBASE-3857
> URL: https://issues.apache.org/jira/browse/HBASE-3857
> Project: HBase
> Issue Type: New Feature
> Affects Versions: 0.90.4
> Reporter: Liyin Tang
> Assignee: Mikhail Bautin
> Attachments: 0001-review_hfile-v2-r1144693_2011-07-15_11_14_44.patch,
> 0001-review_hfile-v2-r1147350_2011-07-26_11_55_59.patch,
> 0001-review_hfile-v2-r1152122-2011_08_01_03_18_00.patch,
> hfile_format_v2_design_draft_0.1.pdf, hfile_format_v2_design_draft_0.3.pdf,
> hfile_format_v2_design_draft_0.4.odt
>
>
> In order to support HBASE-3763 and HBASE-3856, we need to change the format
> of the HFile. The new format proposal is attached here. Thanks for Mikhail
> Bautin for the documentation.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
