[
https://issues.apache.org/jira/browse/HDDS-1735?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16875554#comment-16875554
]
Eric Yang commented on HDDS-1735:
---------------------------------
[~elek]
{quote}Can you please clarify what is your suggestion? How should I improve the
patch?{quote}
The current direction for Ozone continuous integration is not moving in the
direction that I like to see. For example, user can perform:
{code}
mvn clean test
{code}
To perform unit test, and selectively launch test case base on:
{code}
mvn clean test -Dtest=[test-case]
{code}
Integration test can be performed by:
{code}
mvn clean integration-test
{code}
Selective integration test can be performed by:
{code}
mvn clean integration-test -Dtest=[test-case]
{code}
To run in environment specific build use profile:
{code}
mvn clean integration-test -Pk8s
{code}
I am questioning the direction of this issue is good for Ozone continuous
integration because we are creating shell script that wrap around maven
commands and many unique parameters to get Ozone build going. This usually
means the pom.xml and associated configurations are not properly written in
pom.xml and will lead to problems when performing maven release or maven
deploy. This is the reason that I am concerned about introducing more shell
script wrappers.
I really like to have ability to use hadoop-ozone-project, instead of -f
pom.ozone.xml to build the project. This would be Ozone inline similar to
other Hadoop sub-projects. Precommit build can be minimize build time by
building from hadoop-ozone-project. I opened HDDS-1661 for that work. Can we
avoid reference to pom.ozone.xml in the meantime?
{quote}Can you please confirm that this patch doesn't block you in any of your
work? (This is a fix for the existing shell scripts which don't modify any part
of the maven build just help to run certain build steps with maven){quote}
This patch will add code that uses pom.ozone.xml, which I would like to move to
hadoop-ozone-project. It does impact my daily usage of maven, and more effort
to clean up.
> Create separate unit and integration test executor dev-support script
> ---------------------------------------------------------------------
>
> Key: HDDS-1735
> URL: https://issues.apache.org/jira/browse/HDDS-1735
> Project: Hadoop Distributed Data Store
> Issue Type: Improvement
> Reporter: Elek, Marton
> Assignee: Elek, Marton
> Priority: Major
> Labels: pull-request-available
> Time Spent: 1h 10m
> Remaining Estimate: 0h
>
> hadoop-ozone/dev-support/checks directory contains multiple helper script to
> execute different type of testing (findbugs, rat, unit, build).
> They easily define how tests should be executed, with the following contract:
> * The problems should be printed out to the console
> * in case of test failure a non zero exit code should be used
>
> The tests are working well (in fact I have some experiments with executing
> these scripts on k8s and argo where all the shell scripts are executed
> parallel) but we need some update:
> 1. Most important: the unit tests and integration tests can be separated.
> Integration tests are more flaky and it's better to have a way to run only
> the normal unit tests
> 2. As HDDS-1115 introduced a pom.ozone.xml it's better to use them instead
> of the magical "am pl hadoop-ozone-dist" trick--
> 3. To make it possible to run blockade test in containers we should use - T
> flag with docker-compose
> 4. checkstyle violations are printed out to the console
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
