[Fuego] Fuego next status update

Tim.Bird at sony.com Tim.Bird at sony.com
Thu Feb 7 07:17:45 UTC 2019


Hello Fuego users and developers,

Here is a quick status update for Fuego.  Lately I've been working on changes to the batch
processing system.  The new batch_id code seems to work pretty well.  It's in the 'next' branch,
and it needs more testing to figure out the best way to pass the batch_id from parent test
jobs to child test jobs.  But querying runs using batch_id works fine, and this is leading
to an overhaul of the batch system (moving from the testplan json files to Fuego tests
that are themselves collections of sub-tests).  In the process, this will lead to better
facilities to compose complicated test pipelines natively in Fuego (without resorting to
Jenkins pipelines).

Also, I've worked on integrating the 'fserver' code, which is the Fuego
server code for managing job requests between Fuego sites.  This code may end up
supporting other functionality in the future, such as a binary test package cache, and sharing
tests outside of the fuego-core repository.  It's still in the prototype stage, but having it
included by default in the container will allow people to look at it and experiment with it
if they'd like.

Also, I simplified the parser code a bit.  It is now no longer necessary for a test's parser.py
module to preload the python system path.  This is done prior to calling the module, which
is a much cleaner approach, I think.

I've done a bit of refactoring of the directory structure for Fuego.  I've removed the 'engine'
directory in fuego-core, and I've placed fuego-core inside the fuego directory (parallel
to fuego-ro and fuego-rw).  This puts the fuego-core git repository inside the fuego git repository,
which might be a bit confusing.  However, since this configuration matches the directory
structure inside the docker container, it allowed me to remove an ugly hack in ftc that dealt
with this difference.  I also think it's easier for users conceptually.

The downside of this directory refactoring is that patches made against the 'master' branch
will not apply cleanly in the 'next' branch.  Tests that have 'engine' in their paths should
still work, as I've left a compatibility symlink in the repository.  However, managing patches
will be difficult.  For this reason, I'd like to accelerate the 1.5 release, and try to do it much
faster than the 1.4 release.  I'm actually hoping to get it done by the beginning of March.
It will be a much shorter release (only 2 months or so).  I want to shorten the release
window in order to reduce the amount of work spent on patch fixups that are required due
to the directory refactoring.

Finally - I'm moving the main fuego repositories from my personal bitbucket account (tbird20d)
to a 'fuegotest' account.  The new git url will be: https://bitbucket.org/fuegotest/fuego.git
A change to the install process now only requires that you download the 'fuego' repository.
During install, fuego will download 'fuego-core' for you, so this shortens the installation
instructions a bit.

This will all get written up for the 1.5 release notes.  The main reason for this status update
is to let you all know about the shortened release.  If you have a feature pending that you'd like
to get in to this release, please let me know.  And of course, if you are interested in these
features, please try out the 'next' branch (which is now maintained in the fuegotest bitbucket
account) and let me know what you think.

Regards,
 -- Tim




More information about the Fuego mailing list