hadoop-yarn-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Allen Wittenauer ...@altiscale.com>
Subject Re: IMPORTANT: testing patches for branches
Date Thu, 23 Apr 2015 11:21:39 GMT




On Apr 22, 2015, at 11:34 PM, Zheng, Kai <kai.zheng@intel.com> wrote:

> Hi Allen,
> 
> This sounds great. 
> 
>>> Naming a patch foo-HDFS-7285.00.patch should get tested on the HDFS-7285 branch.
> Does it happen locally in developer's machine when running test-patch.sh, or also mean
something in Hadoop Jenkins building when a JIRA becoming patch available? Thanks.


	Both, now that a fix has been committed last night (there was a bug in the Jenkins handling).

	Given a patch name or URL, Jenkins and even running locally will try a few different methods
to figure out which branch to use  out.  Note that a branch name of ‘gitX’ where X is
a valid git reference also works to force a patch to start at a particular commit. 

	For local use, you’ll want to use a ‘spare’ copy of the source tree via the —basedir
option and use the —resetrepo flag.  That will enable Jenkins-like behavior and gives it
permission to make modifications and effectively nuke any changes in the source tree you point
it at.  (Basically the opposite of the —dirty-workspace flag).  If you want to force a branch
(for whatever reason, including where the branch can’t be figured out), you can use the
—branch option. 

	If you don’t use —resetrepo, test-patch.sh will warn that it thinks the wrong branch
is being used but will push on anyway.

	In any case, the result of what it thinks the branch is/should be will be in the summary
output at the bottom along with the git ref that it specifically used for the test.
Mime
View raw message