LAVA development

Understanding the LAVA architecture

The first step for LAVA development is understanding its architecture. The major LAVA components are depicted below:

               +-------------+
               |web interface|
               +-------------+
                      |
                      v
                  +--------+
            +---->|database|
            |     +--------+
            |
+-----------+------[worker]-------------+
|           |                           |
|  +----------------+     +----------+  |
|  |scheduler daemon|---→ |dispatcher|  |
|  +----------------+     +----------+  |
|                              |        |
+------------------------------+--------+
                               |
                               V
                     +-------------------+
                     | device under test |
                     +-------------------+
  • The web interface is responsible for user interaction, including presenting test jobs results, navigating devices, and receiving job submissions through it’s XMLRPC API. It stores all data, including submitted jobs, into the RDBMS.
  • The scheduler daemon is responsible for allocating jobs that were submitted. It works by polling the database, reserving devices to run those jobs, and triggering the dispatcher to actually run the tests.
  • The dispatcher is responsible for actually running the job. It will manage the serial connection to the DUT, image downloads and collecting results etc. When doing local tests or developing new testing features, the dispatcher can usually be run standalone without any of the other components.

On single-server deployments, both the web interface and the worker components (scheduler daemon + dispatcher) run on a same server. You can also install one or more separated worked nodes, that will only run scheduler daemon + dispatcher.

Pre-requisites to start with development

LAVA is written in Python, so you will need to know (or be willing to learn) Python. Likewise, the web interface is a Django application so you will need to Django if you need to modify the web interface.

Also, you will need git.

Contributing Upstream

The best way to protect your investment on LAVA is to contribute your changes back. This way you don’t have to maintain the changes you need by yourself, and you don’t run the risk of LAVA changed in a way that is incompatible with your changes.

Upstream uses Debian, see Developing LAVA on Debian or Ubuntu for more information.

Patch Submissions and workflow

This is a short guide on how to send your patches to LAVA. The LAVA team uses the gerrit code review system to review changes.

So the first step will be logging in to gerrit and uploading you SSH public key there.

If you do not have access to gerrit and it is a small change, you could use another git hosting service, push the LAVA code and point us at commits which could be cherry picked.

Obtaining the repository

There are two main components to LAVA, lava-server and lava-dispatcher.

git clone http://git.linaro.org/git/lava/lava-server.git
cd lava-server

git clone http://git.linaro.org/git/lava/lava-dispatcher.git
cd lava-dispatcher

Setting up git-review

git review -s

Create a topic branch

We recommend never working off the master branch (unless you are a git expert and really know what you are doing). You should create a topic branch for each logically distinct change you work on.

Before you start, make sure your master branch is up to date:

git checkout master
git pull

Now create your topic branch off master:

git checkout -b my-change master

Run the unit tests

Extra dependencies are required to run the tests. On Debian based distributions, you can install lava-dev. (If you only need to run the lava-dispatcher unit tests, you can just install pep8 and python-testscenarios.)

To run the tests, use the ci-run script:

$ ./ci-run

Functional testing

Unit tests cannot replicate all tests required on LAVA code, some tests will need to be run with real devices under test. On Debian based distributions, see Developer package build. See Writing a LAVA test definition for information on writing LAVA test jobs to test particular device functionality.

Make your changes

  • Follow PEP8 style for Python code.
  • Make one commit per logical change.
  • Use one topic branch for each logical change.
  • Include unit tests in the commit of the change being tested.
  • Write good commit messages. Useful reads on that topic:

Re-run the unit tests

Make sure that your changes do not cause any failures in the unit tests:

$ ./ci-run

Wherever possible, always add new unit tests for new code.

Send your commits for review

From each topic branch, just run:

git review

If you have multiple commits in that topic branch, git review will warn you. It’s OK to send multiple commits from the same branch, but note that 1) commits are review and approved individually and 2) later commits will depend on earlier commits, so if a later commit is approved and the one before it is not, the later commit will not be merged until the earlier one is approved.

Submitting a new version of a change

When reviewers make comments on your change, you should amend the original commit to address the comments, and not submit a new change addressing the comments while leaving the original one untouched.

Locally, you can make a separate commit addressing the reviewer comments, it’s not a problem. But before you resubmit your branch for review, you have to rebase your changes against master to end up with a single, enhanced commit. For example:

$ git branch
  master
* my-feature
$ git show-branch master my-feature
! [master] Last commit on master
 ! [my-feature] address revier comments
--
 + [my-feature] address reviewer comments
 + [my-feature^] New feature or bug fix
-- [master] Last commit on master
$ git rebase -i master

git rebase -i will open your $EDITOR and present you with something like this:

pick xxxxxxx New feature or bug fix
pick yyyyyyy address reviewer comments

You want the last commit to be combined with the first and keep the first commit message, so you change pick to fixup ending up with somehting like this:

pick xxxxxxx New feature or bug fix
fixup yyyyyyy address reviewer comments

If you also want to edit the commit message of the first commit to mention something else, change pick to reword and you will have the chance to do that. Just remember to keep the Change-Id unchanged.

NOTE: if you want to abort the rebase, just delete everything, save the file as empty and exit the $EDITOR.

Now save the file and exit your $EDITOR.

In the end, your original commit will be updated with the changes:

$ git show-branch master my-feature
! [master] Last commit on master
 ! [my-feature] New feature or bug fix
--
 + [my-feature] New feature or bug fix
-- [master] Last commit on master

Note that the “New feature or bug fix” commit is now not the same as before since it was modified, so it will have a new hash (zzzzzzz instead of the original xxxxxxx). But as long as the commit message still contains the same Change-Id, gerrit will know it is a new version of a previously submitted change.

Handling your local branches

After placing a few reviews, there will be a number of local branches. To keep the list of local branches under control, the local branches can be easily deleted after the merge. Note: git will warn if the branch has not already been merged when used with the lower case -d option. This is a useful check that you are deleting a merged branch and not an unmerged one, so work with git to help your workflow.

$ git checkout bugfix
$ git rebase master
$ git checkout master
$ git branch -d bugfix

If the final command fails, check the status of the review of the branch. If you are completely sure the branch should still be deleted or if the review of this branch was abandoned, use the -D option instead of -d and repeat the command.

Reviewing changes in clean branches

If you haven’t got a clone handy on the instance to be used for the review, prepare a clone as usual.

Gerrit provides a number of ways to apply the changes to be reviewed, so set up a test branch as usual - always ensuring that the master branch of the clone is up to date before creating the review branch.

$ git checkout master
$ git pull
$ git checkout -b review-111

To pull in the changes in the review already marked for commit in your local branch, use the pull link in the patch set of the review you want to run.

Alternatively, to pull in the changes as plain patches, use the patch` link and pipe that to patch -p1. In this full example, the second patch set of review 159 is applied to the review-159 branch as a patch set.

$ git checkout master
$ git pull
$ git checkout -b review-159
$ git fetch https://review.linaro.org/lava/lava-server refs/changes/59/159/2 && git format-patch -1 --stdout FETCH_HEAD | patch -p1
$ git status

Handle the local branch as normal. If the reviewed change needs modification and a new patch set is added, revert the local change and apply the new patch set.

Adding support for new devices

to LAVA - Board addition howto? Requirements for a device in LAVA

What do I need to create atest image for LAVA? What do I need to create a master image for LAVA? * 8GB SD Card

Writing LAVA extensions

TODO

API Docs

Coming soon.